Chapter  8:  Benign Prostatic Hyperplasia: Diagnosis and Treatment: Clinical Practice Guideline No. 8.

Initial Evaluation

In the initial evaluation of all patients with prostatism, the following are recommended:

• A detailed medical history focusing on the urinary tract, previous surgical procedures, general health issues, and fitness for possible surgical procedures, in order to identify other causes of voiding dysfunction and comorbidities that may complicate treatment. For some patients, a "voiding diary" may help in determining the frequency and nature of the complaints.

• A physical examination, including a digital rectal examination (DRE) and a focused neurologic examination.

• Urinalysis by dipstick testing or microscopic examination of sediment to rule out urinary tract infection and hematuria.

• Measurement of serum creatinine to assess renal function.

In the initial evaluation, measurement of prostate-specific antigen (PSA) is optional. Testing for PSA increases the detection rate for prostate cancer over DRE alone and may detect the cancer at an earlier stage. However, there is significant overlap in PSA values between BPH and prostate cancer patients. As a result, the PSA test does not discriminate well between patients with symptomatic BPH and those with prostate cancer. Also, there is a lack of consensus regarding the evaluation of minimally elevated PSAs, as many of these tests will be "false positives" in patients symptomatic for BPH. This may lead to unnecessary prostate biopsies. In addition, there is no direct evidence as yet whether the increased diagnosis of prostate cancer through PSA testing will lead to a decrease in morbidity and mortality from the disease.

Symptom Assessment

The panel believes that quantification of symptom severity is an important step in the evaluation of men with prostatism.

For objective quantification of symptoms, the recommended instrument is the self-administered American Urological Association (AUA) Symptom Index, which consists of seven questions relating to symptoms of prostatism. Some patients need help understanding and completing such questionnaires. The AUA instrument, however, is generally considered easy to administer and score.

In the AUA scoring system, symptoms are classified as mild (0 to 7), moderate (8 to 19), or severe (20 to 35). The AUA symptom score is recommended for use in treatment planning and periodically in followup.

Additional Diagnostic Tests

Several specific diagnostic tests, discussed below, are available to further assess patients with a presumptive diagnosis of BPH. Data are insufficient to demonstrate the value of these tests for verifying the diagnosis and predicting the results of treatment. Moreover, the optimal thresholds to define normal and abnormal test values are uncertain. Test results thus do not accurately define the severity of BPH. Use of the tests is not mandatory prior to a decision to treat typical patients with a high probability of BPH based on the recommended evaluation. These specific tests, nevertheless, may be valuable if the diagnosis is less certain following the initial evaluation. Other tests may be useful if the patient and physician select more invasive treatment options such as balloon dilation and surgery.

The following tests are optional following the initial evaluation:

• Uroflowmetry is a test that may be useful in patients with symptoms of prostatism because it will identify those whose maximum flowrate is not markedly diminished. These patients are less likely to have bladder outlet obstruction and may respond less well to therapy.

• Pressure-flow studies, while invasive, may be useful in patients whose history and/or examination suggest primary bladder dysfunction (for example, from neurologic disease) as the cause for symptoms of prostatism. Pressure-flow studies may be especially useful in patients for whom a distinction between prostatic obstruction and impaired detrusor contractility might affect the choice of therapy. However, pressure-flow studies may or may not be useful in the workup of the usual patient with symptoms of prostatism.

• Postvoid residual urine (PVR) measurement has not been proven useful in predicting the need for or response to treatment, although patients with large residual urines may have a higher likelihood of failing a watchful waiting treatment strategy. PVR measurements are poorly reproducible for a given patient. However, for patients who elect nonsurgical treatments (including watchful waiting), PVR may be useful in monitoring the course of the disease as PVR measurement may detect worsening obstruction. Repeat measurements should be considered before making treatment decisions based on PVR values. If possible, PVR should be measured noninvasively.

• Urethrocystoscopy is optional during later evaluation if invasive treatment is being planned.

The following tests are not recommended:

• Imaging of the upper urinary tract by ultrasonography or intravenous urography is not recommended for the typical BPH patient. It should be reserved for BPH patients who have concomitant urinary tract disease or abnormalities (for example, hematuria, urinary tract infection, renal insufficiency, previous urinary tract surgery, and history of urinary tract stones).

• Filling cystometry (cystometrography or CMG) is not recommended in the evaluation of typical patients with symptoms of prostatism. It may be useful in the evaluation of patients with suspected neurologic diseases or to determine why individual patients fail to improve after prostate surgery. However, pressure-flow studies are probably a superior alternative in these circumstances.

• Urethrocystoscopy is not recommended as a procedure to determine the need for treatment. Urethrocystoscopy is optional for patients electing invasive therapy (surgery or balloon dilation) in order to assess prostate size and configuration. Urethrocystoscopy may be performed in the office prior to scheduling or in the operating room prior to the procedure. The advantage of the first approach is that the patient can be informed of the results and can participate in the final decision on treatment (for example, surgery as opposed to balloon dilation or transurethral incision of the prostate as opposed to transurethral resection of the prostate). Transabdominal bladder/prostate ultrasonography, but not intravenous urography, can provide the same information on prostate size and configuration noninvasively.

Treatment of BPH

Asymptomatic patients with prostate enlargement due to BPH rarely require treatment. For those BPH patients who have specific complications due to BPH, such as urinary retention, prostate surgery is usually the most appropriate form of treatment (see Decision Diagram, Attachment A). All other patients should, in consultation with the physician, decide on the treatment after considering the possible outcomes of each potential treatment.

The depth of information desired will vary from patient to patient. To make a treatment decision, the patient needs to consider how "bothered" he is by his symptoms as well as his attitude toward the possible benefits and risks of each treatment option. The health care provider can help guide the patient in making the most appropriate treatment decision.

Complications of BPH

Little is known about the ultimate complications and morbidity associated with untreated BPH. Internationally, between 20 and 50 percent of patients are in urinary retention when they undergo prostate surgery. However, the likelihood that a patient with a given symptom complex will go into complete urinary retention over a given timespan is not well defined.

Even less is known about the cumulative risk of obstructive uropathy and chronic renal failure among men with BPH. Patients with histologic hyperplasia and urodynamic obstruction but without bothersome symptoms are said to have "silent" BPH. Although this condition is probably rare, it does occur (Mukamel, Nissenkorn, Boner, et al., 1979 ; Powell, Smith, and Feneley, 1980 ; George, Feneley, and Roberts, 1986 ; Nissenkorn, Savion, and Servadio, 1988 ; Parys, Machin, Woolfenden, et al., 1988 ; Sacks, Aparicio, Bevan, et al., 1989 ; Sarmina, and Resnick, 1989 ; Ghose, 1990). It is difficult to identify a clinical indicator that might be useful in predicting which patients will develop serious BPH complications without bothersome symptoms. Only large-scale screening efforts could identify those few patients with silent prostatism.

A small fraction of patients have urinary tract infections when they are treated for BPH, but it is unknown what risk factors predispose men to this complication. Also unknown is the extent to which a urinary tract infection in men with BPH predisposes them to renal function deterioration.

Bladder decompensation as a result of chronic outlet obstruction may be a long-term complication of untreated BPH in some patients. However, the probability of detrusor muscle decompensation, particularly decompensation not reversible by surgery, in men who are following a strategy of watchful waiting is unknown.

Bladder stones may occur as a result of bladder outlet obstruction, urinary tract infection, a retained foreign body (such as a fragment of a catheter balloon), and dietary factors. The prevalence of bladder stones in men who undergo surgery for BPH is approximately 1 to 2 percent. The incidence of bladder stones over time in men with symptomatic BPH is unknown.

Mortality

Accurate data on BPH as a primary cause of death are currently not available for most countries in the world. Mortality ascribed to BPH varies from region to region and from country to country. Ekman (1989) has attempted to use World Health Organization death rates for BPH from 50 countries, presenting the death rates as the number of deaths per 100,000 males age 45 and older. The reported death rates ranged from 29.7 per 100,000 men over 45 years in Eastern Germany (1984) to 0.5 per 100,000 males older than 45 years in Singapore (1985). The reported rate for the United States is 1.8 per 100,000 (1983). However, lack of uniformity in criteria for attributing deaths to BPH limits the usefulness of these data.

Quality of Life

Although mortality from BPH is extremely rare in the United States and serious complications are uncommon, the symptoms of prostatism are bothersome to many patients. If symptoms interrupt normal daytime activities or sleep, create anxiety, or reduce perception of general health, the quality of life can decline significantly. Yet, the degree of "bothersomeness" may be dramatically different for different patients with the same degree of symptom severity (Barry, Fowler, O'Leary, et al., 1992a). Because BPH is primarily a quality-of-life disease, and because significant variations do exist in individual patients' perceptions of bothersomeness and treatment risk, the BPH Guideline Panel concluded that the patient should play a central role in determining the need for treatment. A caring health care provider and patient can work together to reach an optimal treatment decision.

Practice Variation

Rates of prostatectomies performed per population unit vary greatly among different countries due to differences in health care delivery systems. Rates of prostatectomies performed also vary greatly between small geographic areas, including areas within the United States (McPherson, Wennberg, Hovind, et al., 1982 ; Wennberg, Mulley, Hanley, et al., 1988). The rate of prostatectomies may vary as much as fourfold between small geographic areas (Barry, 1991). These variations are unlikely to be due to geographic differences in disease prevalence or severity and may often be the result of variations in decisionmaking by individual practitioners.

The variation in physician decisionmaking is significant regarding indications for BPH treatment. As noted above, this variation exists in part due to the uncertainty in the medical literature regarding the natural history of BPH and its treatment outcomes. Also, current BPH practice policies, developed by individual State peer review organizations (PROs) to assess the "appropriateness" of surgery for reimbursement purposes, are highly variable between States.

Overall, the number of procedures being performed is high. It is estimated that one in every four men in the United States will undergo treatment for the relief of symptomatic BPH by the age of 80 (Barry, 1990, 1991). In 1990, approximately 335,000 surgical procedures for BPH -- most commonly transurethral resection of the prostate (TURP) -- were performed in the United States. TURP is the second most common surgical procedure performed in Medicare-age men, surpassed only by cataract surgery. The resulting total cost is an estimated $2 billion to $3 billion per year (Holtgrewe, Mebust, Dowd, et al., 1989).

A survey of American urologists revealed that TURP represents 38 percent of the major surgical procedures they performed and 24 percent of their total work load (Holtgrewe, Mebust, Dowd, et al., 1989). The surgical fee for transurethral prostatectomy in this survey ranged from $998 in the southern United States to $1,829 in the Northeast. The number of TURPs performed by each urologist dropped from 101 in 1962 (number of respondents = 266) to 67 in 1986 (number of respondents = 219).

Demographic trends may affect these numbers. The population of American males over 65 years of age will increase significantly by the year 2000. The impact of an aging population on the cost of BPH treatment is obvious. The rate of prostatectomies increases dramatically with increasing age (Glynn, Campion, Bouchard, et al., 1985 ; Arrighi, Guess, Metter, et al., 1990).

Although surgical costs are significant, the growing number of nonsurgical therapies for BPH may indirectly increase costs further because of additional retreatment costs. Some men will choose nonsurgical therapies instead of surgery initially, but some will still eventually undergo TURP. Moreover, the availability of treatments with fewer side effects may draw many more men into the medical system for treatment. This trend has been bolstered by advertising campaigns designed to swell the numbers of men seeking treatment. An investment group (Shearson, Lehman, and Hutton) predicts that the percentage of the eligible population being treated for BPH might increase from 15 to 70 percent between the years 1990 and 1995, expanding the total number of patients treated per year from 450,000 to 2.1 million (McCarthy, and Willard, 1990). However, it is suggested by the same group that the percentage of patients treated surgically will drop to 40 percent, while the percentage of patients treated with presumably less expensive drugs and other devices will increase accordingly. This projection may not be accurate, but it clearly demonstrates that industrial interests will seek to influence the decisions of patients and health care providers.

The high prevalence of BPH in the Medicare population, the unexplained geographic variations in practice patterns, and the high cost of treatment all make BPH an appropriate candidate for an evidence-based guideline.

In developing guideline recommendations for the diagnosis and treatment of BPH, the panel members were themselves guided throughout their deliberations by the idea that the best interest of the patient should be the highest goal of any health care provider. The basic motto "Primum Non Nocere" ("Above all else, do no harm") is especially relevant for evaluation and treatment of a disease that is very rarely lethal and has widely varying effects on the quality of life of the men afflicted by it.

Literature Review

The BPH Guideline Panel conducted an extensive literature review on BPH, utilizing the Cumulated Index Medicus and MEDLINE data bases. The comprehensive retrieval of published manuscripts and relevant unpublished material yielded approximately 1,200 abstracts for review. More than 200 manuscripts were chosen for indepth analysis. The panel used only articles published in peer-reviewed, mostly English-language, journals. Excluded as a rule were articles in non-peer-reviewed journals, book chapters, advertisements, and technical and promotional information from device manufacturers or pharmaceutical makers. Any exception was documented with a rationale for the exception. In addition, the panel sought available evidence on the costs of BPH and its treatment. Published reports and expert opinions helped provide insight into the problem.

Analysis Methodology

The BPH Guideline Panel adapted an explicit approach to designing practice policies as outlined by David M. Eddy, MD, PhD (1992). In this approach, the specific data, subjective judgments, patient preferences, and techniques utilized in generation of the policy are explicitly articulated.

The major limitation of an explicit approach to a BPH practice policy is the paucity of well-designed clinical research studies available in the literature. The small number of randomized clinical trials of treatments and the lack of information concerning the sensitivity, specificity, and predictive value of diagnostic tests significantly limit the quality and strength of the guidelines generated. However, the use of an explicit approach also led to the development of a clinical research agenda to address this problem.

Syntheses of the results from the structured literature reviews utilized the confidence profile method developed by Eddy and Hasselblad (Eddy, 1989 ; Eddy, Hasselblad, and Shachter, 1990 ; Eddy, and Hasselblad, 1992). Analyses of combined data from a variety of treatment studies were performed using the confidence profile method software package FAST*PRO, which allows the calculation of confidence profiles (or confidence intervals [CIs]) around the mean or median probability for a given outcome to occur (Eddy, and Hasselblad, 1992).

The confidence profile method is based on Bayesian statistical theory. The combination of evidence is performed using either Bayes' formula or the hierarchical Bayes' formula (Adar, Critchfield, and Eddy, 1989 ; Eddy, 1989 ; Eddy, Hasselblad, and Shachter, 1990). Unless otherwise indicated, the hierarchical Bayes' formula was used for all outcome analyses presented in this Clinical Practice Guideline.

The panel elected to report the results of the mathematical combination of evidence as a 90-percent CI (5th to 95th percentile) to indicate numerically the range of uncertainty about the probability of a given outcome. For example, if the 90-percent CI for the combined evidence for a particular treatment outcome ranges from 69 to 74 percent (90-percent CI 69-74 percent), it is known that for every 1,000 such confidence intervals constructed, about 900 of them will contain the true value of this outcome.

Every outcome the panel thought important to patients receiving treatment for BPH was analyzed using the confidence profile method software FAST*PRO. Evidence from all studies reporting the outcome was combined for each treatment modality. The data obtained from the analyses were used for the patient preference survey, as these data currently represent the panel's best estimates of the relevant outcome probabilities.

Guideline Terms

An important factor in the use of clinical practice guidelines is their degree of flexibility (Eddy, 1990a, 1992). Terms utilized in this Clinical Practice Guideline to denote degree of flexibility in the clinical decisionmaking process are described below.

In reference to diagnostic tests, the panel utilized the terms recommended, optional, and not recommended to indicate desirability of specific diagnostic interventions. If a test was categorized as optional, there is clear evidence of its benefit for certain patients. However, the data are insufficient to demonstrate the test's value for routine patients in confirming the diagnosis of BPH and predicting the results of treatment, and the definition of normal and abnormal test values may be uncertain. The evidence is thus insufficient to mandate use of the test prior to a decision to treat.

If a test was not recommended, the panel believed there is sufficient evidence to indicate that the test does not have value or has potential harms that exceed its potential benefits in routine cases.

With regard to treatment policies, the panel utilized the terms standard, guideline, and option to denote intended degree of flexibility. The meaning of each term is explained below.

A treatment policy is considered a standard if the health and economic consequences of the intervention are sufficiently well-known to permit a conclusion that the intervention is clearly superior to all alternatives. In addition, a standard requires virtual unanimity among patients about the desirability or undesirability of the intervention and knowledge of the potential outcomes of both the treatment and the alternatives. Virtual unanimity implies that at least 95 percent of the people asked agree on overall desirability of the treatment, relative to the alternatives.

A treatment policy is considered a guideline if the outcomes of the intervention are well-enough understood to permit a conclusion that the treatment is probably superior (or inferior) to all alternatives and is preferred (or not preferred) by an appreciable, but not unanimous majority of patients (60 to 95 percent of people agreeing on overall desirability, and proper use). Depending on the patient, the setting, and other factors, guidelines can and should be tailored to fit individual needs.

A treatment policy is called an option if (1) the outcomes are not known with certainty, (2) the outcomes are known but patient preferences are not, or (3) patient preferences are evenly divided (40 to 60 percent of people preferring the intervention, with the remainder preferring an alternative). Even division can represent indifference on the part of patients or a wide but even division of preferences. For different reasons, all options are intended to be flexible. Any policy could therefore be an option with outcome uncertainty, an option with preference uncertainty, an option with indifference in regard to preferences, or an option with division of preferences.

Background

In the early 1970s, reports identifying prostate-specific antigens were first published by Ablin, Soans, Bronson, et al. (1970) and Hara, Inorre, and Fukuyama, (1971). The same antigen was isolated from prostatic tissue, purified, and demonstrated to be specific for prostatic tissue (Wang, Valenzuela, Murphy, et al., 1979-80). Although identified in all types of prostatic tissue (normal, benign hyperplastic, and malignant), it could not be found in any other human tissue. Thus, it became known as prostate-specific antigen.

PSA is a glycoprotein with kallikrein-like serine protease activity produced by the epithelial cells that line the acini and ducts of the prostate gland. PSA is secreted into the prostatic ductal system and causes liquefaction of the seminal coagulum at the time of ejaculation.

Concentrations of PSA per gram of tissue do not differ significantly among normal, hyperplastic, and cancerous prostate tissue. PSA can be detected in the serum of young men with small, nonhyperplastic prostates, older men with BPH, and men with localized or metastatic prostate cancer. For the individual with no prostatic disease (prostatitis, BPH, or prostate cancer), the serum concentration is extremely low. It has recently been demonstrated that the mean serum value for such a patient is 0.07+/-0.4 ng/mL (Glenski, Klee, Bergstrahl, et al., 1992). This is markedly lower than the upper limit of the reference range (4.0 ng/mL) by either the Tandem-R PSA assay or the IMx PSA assay (Oesterling, 1991).

For PSA to enter the general circulation, it must pass through a number of physiologic barriers, including the basal cell layer, the basement membrane of the acini, the prostatic stroma, the capillary basement membrane, and the capillary endothelial cells. Breakdown of these barriers, allowing serum PSA concentration to increase, may be caused by a number of conditions. They include prostatic infarction (a common phenomenon associated with acute urinary retention), prostatitis, BPH, and prostate cancer. Thus, an elevated serum PSA level is not pathognomonic for prostate cancer.

Agents and Procedures Affecting PSA Levels

The serum PSA level is affected by the new drug finasteride. Finasteride therapy for BPH reduces serum PSA by approximately 50 percent. Whether longitudinal followup of the serum PSA will have, for men on finasteride, the same diagnostic value as in the general population remains unclear. (Theoretically, differential suppression of PSA in men with BPH and prostate cancer could actually increase the information value of PSA measurements after treatment with finasteride.) If the patient and the physician elect to include PSA measurement in the initial BPH evaluation, a baseline should be reestablished after beginning finasteride.

Prostatic biopsy, surgery, and urethrocystoscopy can temporarily elevate the serum PSA level. It had been speculated that manipulation of the prostate during DRE might also raise the serum PSA level. However, in a multicenter study of 2,754 healthy men age 40 years and older (Crawford, Schutz, Clejan, et al., 1992), the increase observed after a DRE was not clinically or statistically significant for men with a serum PSA level <10.0 ng/mL (n = 2,667). For men with PSA values >10.0 ng/mL, a trend toward increased serum PSA levels was noted. At these already high levels, the increase would not interfere with clinical management.

Chybowski, Bergstralh, and Oesterling (1992), in the only prospective randomized study examining this issue, also found that a DRE does not have a clinically significant effect on serum PSA concentration. The median increase following a DRE was 0.4 ng/mL. For the majority of patients, such a minimal increase would not alter clinical management. Unlike for prostatic acid phosphatase, physicians do not have to obtain a serum PSA level before performing a DRE. Neither do they have to wait a period of time after the DRE to allow the serum PSA concentration to return to baseline before measuring the serum PSA level.

Serum PSA Levels and BPH

Table 1. Serum PSA concentration in patients with histologically (more...)

Table 1. Serum PSA concentration in patients with histologically confirmed BPH [1]

Reference	Number of patients	Percentage and number of patients with PSA in specified range[2] (ng/mL)
		0.0-4.0	4.1-10.0	> 10.1
Ercole, Lange, Mathisen, et al., 1987	357	79% (282)	18% (64)	3% (11)
Oesterling, Chan, Epstein, et al., 1988	72	47% (34)	43% (31)	10% (7)
Armitage, Cooper, Newling, et al., 1988	91	53% (48)	35% (32)	12% (11)
Hudson, Bahnson, and Catalona, 1989	168	79% (133)	19% (32)	2% (3)
Total	688	72% (497)	23% (159)	5% (32)

1. Tandem-R PSA assay was used.

2. Number of patients is in parentheses.

Other studies confirm the association between BPH and elevated serum PSA levels. Stamey, Yang, Hay, et al. (1987) used the Pros-Check PSA assay (Yang Laboratories, Bellevue, WA) to evaluate 73 patients with BPH and found 86 percent to have an elevated serum PSA level. The preoperative values ranged from 0.3 to 37 ng/mL, with the mean level being 7.9 ng/mL. After TURP, the mean value decreased to 1.3 ng/mL, with the range being 0.1 to 6.7 ng/mL. Based on these findings, the investigators concluded that benign hyperplastic tissue elevates the serum PSA level at a rate of 0.3 ng/mL per gram of BPH tissue (0.2 ng/mL by the Tandem-R PSA assay).

Daver, Soret, Coblentz, et al. (1988) found elevated PSA levels in 68 percent and 70 percent, respectively, of clinically and histologically confirmed BPH patients (n = 150). Another study (Buamah, Johnson, and Skillen, 1988), using the Tandem-R PSA assay with an established upper limit of normal of 5 ng/mL, found that 21 of 45 patients (47 percent) with histologically proven BPH had an elevated serum PSA level. Filella, Molina, Jo, et al. (1990), also utilizing the Tandem-R PSA assay, found that 87 percent of patients with BPH had PSA levels greater than 2 ng/mL; 13 percent had levels greater than 10 ng/mL.

Distinguishing Prostate Cancer from BPH

To be a valuable detector of early prostate cancer in patients with symptoms of prostatism, a PSA test must be able to identify and distinguish curable cancer from purely benign conditions of the prostate. In other words, a PSA test must have high sensitivity and specificity, low false-negative and false-positive rates, and high negative and positive predictive values.

Table 2. Comparative PSA values for BPH and organ-confined (more...)

Table 2. Comparative PSA values for BPH and organ-confined prostate cancer [1, 2]

Reference	Percentage and number of patients with PSA in specified range[3] (ng/mL)
	0.0-4.0		4.1-10.0		>= 10.1
	BPH	Prostate Cancer	BPH	Prostate Cancer	BPH	Prostate Cancer
Partin, Carter, Chan, et al., 1990	47%	45%	46%	44%	7%	11%
	(34/72)	(83/18 )	(33/72)	(82/18 )	(5/72)	(20/18 )
Lange, Ercole, Lightner, et al., 1989	79%	45%	18%	32%	3%	23%
	(282/3 7)	(14/31)	(64/35 )	(10/32)	(11/35 )	(7/31)
Hudson, Bahnson, and Catalona, 1989	79%	38%	19%	26%	2%	36%
	(133/1 8)	(39/10 )	(32/16 )	(27/10 )	(3/168)	(37/103 )
Total	75%	43%	22%	37%	3%	20%
	(449/5 7)	(136/3 9)	(129/5 7)	(119/3 9)	(19/59 )	(64/31 )
p value	p < 0.0001		p < 0.001		p < 0.0001

1. Both diseases (BPH and organ-confined prostate cancer) were confirmed histologically.

2. Tandem-R PSA assay was used.

3. First number in parentheses is number of patients in specified PSA range. Second number is total number of patients with disease.

Table 3. Ability of PSA to distinguish organ-confined (more...)

Table 3. Ability of PSA to distinguish organ-confined prostate cancer from BPH, based on data in Table 2

Test parameter	PSA values >= 4 ng/mL	PSA values >= 10 ng/mL
Sensitivity	57%	20%
Specificity	75%	97%
False negative rate	43%	80%
False positive rate	25%	3%

Comparing the individual serum PSA values for men who have BPH with the values for patients who have organ-confined prostate cancer, Partin, Carter, Chan, et al. (1990) found no statistically significant difference between the two groups. The mean value (+/- standard error) for men with BPH was 5.98+/-1.0 ng/mL, whereas the mean level for the patients with organ-confined prostate cancer was 5.62+/-0.6 ng/mL. Similar results are obtained when clinical stage A and B prostate cancer patients are compared with men who have BPH (Chan, Bruzek, Oesterling, et al., 1987). These findings indicate that serum PSA concentration does not have either the sensitivity or the specificity of an ideal detector. On an individual basis, serum PSA is not a reliable test for distinguishing men with BPH from patients with early prostate cancer.

Serum PSA Compared with DRE

Recently, two large-scale studies have been conducted examining the role of PSA as a screening test for prostate cancer (Catalona, Smith, Ratliff, et al., 1991 ; Brawer, Chetner, Beatie, et al., 1992). At Washington University in St. Louis, Missouri, Catalona and associates evaluated 1,653 healthy, asymptomatic men who had a serum PSA concentration; 37 of the men were diagnosed with cancer, for an overall detection rate of 2.2 percent. At the University of Washington in Seattle, Brawer and colleagues had a detection rate of 2.6 percent in a study of 1,249 men. These values are somewhat better than the 1.3 to 1.5 percent reported for DRE by Cupp and Oesterling (1993).

In the St. Louis study, 32 percent of the cancers identified would have been missed if DRE alone had been used. The Seattle group found that 38 percent of cancers in their study population would not have been identified if DRE alone had been used. From these studies, it appears that measurement of the PSA level can identify prostate cancer not detectable by DRE (impalpable lesions) and that, for a given population, PSA measurements can identify 50 to 61 percent more prostate cancers than can DRE.

However, DRE has also been shown to detect prostate cancers not identifiable by an elevated serum PSA concentration. Catalona, Smith, Ratliff, et al. (1991) examined 300 men who underwent prostate biopsy for various reasons, but not necessarily because of an elevated serum PSA value. They found that 21 percent of the patients with biopsy-proven cancer had a serum PSA value in the reference range (0.0 to 3.9 ng/mL, using the Tandem-R PSA assay) and would have been missed if the decision to take a biopsy specimen were made on the basis of PSA level alone. Cooner, Mosley, Rutherford, et al. (1990), evaluating over 1,800 patients presenting to a urologic practice, made very similar observations.

These data suggest (1) that PSA measurements cannot identify all detectable cancers and (2) that DRE can detect lesions that would be missed by relying on the serum PSA concentration alone. All the studies taken together indicate that the most sensitive test strategy for prostate cancer is use of both DRE and serum PSA measurements. Some cancers are identified by DRE, some by serum PSA. They are not necessarily the same lesions.

PSA Velocity and PSA Density

The ability of serum PSA measurement to detect prostate cancer at an early, curable stage and distinguish it from BPH was examined further by Carter, Pearson, Metter, et al. (1992). Carter and associates investigated the concept of "PSA velocity," the rate of change in serum PSA. In a unique study based on banked serum from the Baltimore Longitudinal Study of Aging (BLSA), they demonstrated that the rate of change was more useful than the actual serum PSA level for detecting prostate cancer. When a cutoff of 0.75 ng/mL or higher per year for the rate of change was used, the specificity was 90 percent as compared with 60 percent for the cutoff of 4.0 ng/mL or higher for the serum PSA concentration. The sensitivity for the rate of change, however, was not significantly better than for the serum PSA concentration. The number of cancer patients studied was small (n = 20). The validity of PSA velocity needs further prospective evaluation.

Another technique for improving PSA's ability to detect early prostate cancer is to correlate the serum PSA concentration with the prostatic volume. A mildly elevated serum PSA level associated with a small prostate gland may be indicative of cancer, whereas the same value in a patient with a large gland may be only indicative of BPH.

To investigate this concept, Benson and associates (Benson, Whang, Olsson, et al., 1992 ; Benson, Whang, Pantuck, et al., 1992) defined the parameter of "PSA density." PSA density is a quotient, the serum PSA concentration divided by the volume of the prostate gland as determined by transrectal ultrasonography. The investigators found this parameter useful for distinguishing men with BPH from patients with prostate cancer, especially when the serum PSA value is between 4 ng/mL and 10 ng/mL and the DRE is negative. Patients with values in this "mildly elevated range" who also have an elevated PSA density may be at increased risk for having prostate cancer. Patients with PSA values in the same range, but with a low PSA density, are unlikely to harbor a malignancy in the prostate.

PSA density may become a useful tool to help physicians decide which patients with an intermediate PSA level should undergo prostate biopsy and which to follow with annual evaluation. Other studies are underway. Currently, however, the panel is unable to validate the predictive value of PSA density. Like PSA velocity, PSA density requires further investigation before it can be recommended as a valid diagnostic test.

Impact of PSA Testing on Ultimate Outcomes

When the role of PSA in the diagnostic workup of patients with BPH is evaluated, a key question is how many of the approximately 10 to 15 percent of patients found to have stage A1 or A2 prostate cancer at the time of TURP would be detected prior to surgery. Some of these men might be approached differently if a preoperative evaluation uncovered an impalpable prostate cancer.

Although most of this discussion has focused on the accuracy of PSA testing, a more important issue is that the value of early prostate cancer detection itself is controversial. Unfortunately, no data are available to establish that earlier diagnosis leads to decreased morbidity or mortality. The average age of patients undergoing surgical treatment of BPH is 67 years. Therefore a substantial number of men presenting with prostatism will be over 70, when the value of aggressive treatment with radical prostatectomy or radiation therapy is subject to question even by advocates of early detection and treatment.

Several large-scale investigations are underway to address these issues. Until this information becomes available, it is reasonable that a serum PSA determination be considered an optional diagnostic test in evaluating men presenting with prostatism. PSA determination would presumably be of most value in those men for whom the diagnosis of nonpalpable prostate cancer would change the BPH treatment recommendation.

Procedure

The procedure for uroflowmetry has been standardized. The patient should be instructed prior to the test, but not distracted by verbal comments or observation during the actual flowrate recording. Flowrate should be measured with the patient in his normal voiding position (supine, sitting, or standing), in private, and when he feels the urge to void. Although bladder filling is needed, the patient should not be encouraged to wait until he cannot hold his urine any longer or until the bladder is severely distended. It is helpful to inquire whether the patient judged the voiding act normal according to his own habits. If doubts exist whether the voiding was representative of the patient's normal voiding, the test should be repeated and the first measurement disregarded.

Measured Variables

According to the International Continence Society (ICS) Committee on Standardization of Terminology (Abrams, Blaivas, Stanton, et al., 1988), urinary flow may be described in terms of rate and continuous or intermittent pattern. Flowrate is defined as the volume of fluid expelled via the urethra per time unit and is expressed in milliliters per second (mL/sec). Voided volume, environment, patient position, the mode of filling (diuresis or catheter), and the type of fluid used should be specified.

Voided volume is defined as the total volume expelled via the urethra. Maximum or peak flowrate (Q[max]) is the maximum measured flowrate. Average flowrate (Q[ave]) is the voided volume divided by flowtime (meaningful only if flow is continuous), and flowtime is the time during which measurable flow occurs. In the case of intermittent flow, the intervals between flow episodes are disregarded. The parameter of voiding time (total duration of micturition) is reported.

One aspect of interpreting flowrate recordings involves discrepancies between manual and automatic (by the machine) readings. Grino, Bruskewitz, Blaivas, et al. (1992) analyzed a data base of 1,645 patients who voided a total of 23,857 times. The manually read Q[max] values were on average 1.5 mL/sec lower than the automatic machine-read values. Artifacts of >2 mL/sec were found in 20 percent of tracings, and artifacts of >3 mL/sec were found in 9 percent of tracings. The manually read peak flowrate resulted in CIs 15 to 25 percent smaller than for the machine-read values, thus improving the statistical power of peak flowrate recordings when treatment groups are compared.

Correlation Between Flowrate and Voided Volume

An important issue is whether a flowrate measurement can be interpreted without considering the voided volume. Many investigators have studied the dependence of Q[max] on voided volume and have developed nomograms reflecting this relationship (von Garrelts, 1956, 1957, 1958 ; Scott, and McIlhaney, 1961 ; Beck, and Gaudin, 1969 ; Susset, Picker, Kretz, et al., 1973 ; Drach, Layton, and Binard, 1979 ; Drach, and Steinbronn, 1986 ; Siroky, Olsson, and Krane, 1979, 1980 ; Layton, and Drach, 1981 ; Ryall, and Marshall, 1982 ; Balslev-Jorgensen, Jensen, Bille-Brahe, et al., 1986 ; Haylen, Ashby, Sutherst, et al., 1989). There is no universally accepted nomogram for flowrate correction.

Figure 2. Relationship between Q[max] and voided volume (more...)

   Figure 2. Relationship between Q[max] and voided volume

Figure 2

In addition, the panel reviewed data submitted to the FDA from the phase III finasteride trials (Merck, Sharp & Dohme Research Laboratories, 1992). The data demonstrate minimal Q[max] volume dependence in men with BPH if the voided volume is >125 mL.

Although Q[max] partially depends on the voided volume and age, the range of volumes for which this correlation can be expressed as a simple mathematical formula or a nomogram is uncertain. The nomograms themselves vary considerably. Furthermore, Q[max] may not be as volume dependent in men with BPH as it is in normal men. Therefore, no recommendation can be made for a single volume correction technique. If volume correction is not utilized, the inaccuracy of flowrates with voided volumes of <125 to 150 mL should be recognized.

Sensitivity and Specificity

The crucial issues for flowrate recording in regard to patients with symptoms of prostatism are: (1) the test's sensitivity in detecting patients with BPH, (2) the test's specificity in excluding those with symptoms due to some other cause than BPH, and (3) the test's ability to identify those patients most likely to have a positive outcome following treatment and those patients who most likely will not have a positive outcome.

Test Specificity.

Table 6. Inability of eight urodynamic parameters to (more...)

Table 6. Inability of eight urodynamic parameters to separate obstruction from detrusor decompensation

Parameters	Obstruction (n = 31)	Detrusor failure (n = 14)
Time to Qmax (sec)	7.6 ± 4.3	11 ± 8.9
Voiding time (sec)	41 ± 20	65 ± 62
Qmax (mL/sec)	8.5 ± 2.1	8.8 ± 2.7
Qave (mL/sec)	4.4 ± 1.4	4.3 ± 1.7
Qmax/time to Qmax	1.3 ± 0.75	0.89 ± 0.64
Time from Qmax to 95% of volume voided/time to void initial 95% of volume	0.76 ± 0.13	0.66 ± 0.21
Time to Qmax/total voiding time	0.2 ± 0.11	0.28 ± 0.18
Qmax/Qave	2.0 ± 0.62	2.1 ± 0.5

Note: Qmax = peak flow rate. Qave = average flow rate.

Source: Chancellor, Blaivas, Kaplan, et al., 1991.

Chancellor, Blaivas, Kaplan, et al. (1991) found that flowrate recording cannot distinguish between bladder outlet obstruction and impaired detrusor contractility as the cause for a low Q[max]. None of 8 measured, noninvasive urodynamic parameters was significantly different for 31 patients with outlet obstruction than for 14 patients with impaired detrusor contractility (Table 6).

Predictive Value of Uroflow

Abrams (1977) and Abrams, Farrar, Turner-Warwick, et al. (1979) studied the value of uroflowmetry prior to prostatectomy. Failure rates for surgery were found to decrease with the addition of flowrate measurement to symptom assessment in preoperative evaluation.

Q[max] appears to predict surgical outcome in some studies. In one study (Jensen, Bruskewitz, Iversen, et al., 1984), 53 patients underwent prostatectomy based on clinical indication alone. The study population was divided according to Q[max] into three groups: Q[max] less than 10 mL/sec, between 10 and 15 mL/sec, and greater than 15 mL/sec. All three groups experienced improvements in their symptom score after surgery, but the group with a Q[max] less than 10 mL/sec before treatment had a better overall subjective outcome as assessed by global subjective judgment.

Table 7. Symptomatic improvement based on pretreatment (more...)

Table 7. Symptomatic improvement based on pretreatment Qmax; statistically significant difference in probability for treatment success below or above Qmax cutoff value of 15 mL/sec

Symptomatic Improvement	Qmax > 15 mL/sec	Qmax > 10 and < 15 mL/sec	Qmax < 10 mL/sec
Much better	12 patients	31 patients	76 patients
No change	5 patients	3 patients	7 patients
Success rate	71%	91%	92%
Success rate if Qmax > 10 as cutoff = 84%
Success rate if Qmax < 15 as cutoff = 92%

Note: Qmax = peak flowrate.

Source: Jensen, Jorgensen, and Mogensen, 1988a.

Table 8. Diagnostic specificity and sensitivity of (more...)

Table 8. Diagnostic specificity and sensitivity of flowrate recording using different Qmax cutoff values

Qmax cutoff value	Specificity = p(S|Q)[1]	Sensitivity = p(!S|Q)[2]
10	92%	16%
15	91%	29%

1. p(S|Q) = probability for successful outcome (S) if Qmax ≥ indicated value.

2. p(!S|Q) = probability for unsuccessful outcome (!S) if Qmax ≥ indicated value.

Note: Qmax = peak flowrate.

Source: Jensen, Jorgensen, and Mogensen, 1988a.

Unfortunately, not all investigators have utilized 15 mL/sec as the cutoff. McLoughlin, Gill, Abel, et al. (1990), using urodynamic testing and a cutoff of 12 mL/sec, evaluated 108 men with prostatism before and 1 year after surgery and derived the following conclusions:

1. A Q[max] less than 12 mL/sec as indicator for obstruction would subject only 3 percent of patients unnecessarily to TURP.

2. Routine pressure-flow studies or cystometrograms are not indicated, but the screening of flowrates followed by further urodynamic testing in patients with a Q[max] of greater than 12 mL/sec is recommended. Studies to rule out uninhibited detrusor contractions are also indicated if symptoms persist after treatment.

3. The data confirm the independent predictive value of symptoms and flowrate.

Very low flowrates do not appear to portend poor treatment outcome. In one study (Dorflinger, Bruskewitz, Jensen, et al., 1986) of 84 patients undergoing surgery for symptomatic BPH, patients with a low preoperative Q[max] (less than 7 mL/sec) improved symptomatically as much as patients with a Q[max] greater than 7 mL/sec.

Conclusions

Neither subjectively assessed symptoms nor quantified symptom-score analysis correlate strongly with uroflowmetry measurements. Both are independent assessment tools. Patients with a peak flowrate (Q[max]) greater than 15 mL/sec have somewhat poorer outcomes than those with a Q[max] less than 15 mL/sec (although the majority of patients still improve). Other investigators report similar findings for different Q[max] cutoff values (for example, 12 mL/sec). Patients with very bothersome symptoms suggestive of prostatism, but having a Q[max] greater than 15 mL/sec, may benefit from further urodynamic testing (that is, pressure-flow studies) to reduce the number of surgical treatment failures. A Q[max] less than 15 mL/sec does not differentiate between outflow obstruction and detrusor impairment. Apparently, no minimal threshold of Q[max] reliably diagnoses detrusor failure or predicts a poor surgical outcome. The wide range of Q[max] "cutoffs" used by State PROs is not surprising. Every flowrate threshold used can be supported by at least one small study.

Until long-term outcome studies clearly establish the role of flowrate measurements, their use should be optional.

Methods of Measurement

PVR measurement can be performed by noninvasive (ultrasound) and by invasive (catheterization) methods. The most common method is urethral in-and-out catheterization. Invasive techniques are accurate if performed correctly, but carry a small risk of discomfort, urethral injury, urinary tract infection, and transient bacteremia (which has not been quantified in the literature).

Although in-and-out catheterization is accepted as the "gold standard" for measuring PVR, two studies note technical problems with this technique. Stoller and Millard (1989) assessed the accuracy of PVR measurement with in-and-out catheterization done by urological nurses in 515 patients. In 26 percent of patients, the bladder was not empty after catheterization. In an initial group of 116 male patients, 30 percent had an inaccurate assessment. The mean difference in this group between the initial and the actual residual volume was 76 mL. In another group of 106 male patients, only 14 percent (n = 15) had inaccurate initial measurements with a mean difference of 85 mL. This second group was investigated after the nurses were educated regarding techniques to improve complete draining of the bladder.

Purkiss (1990) evaluated 15 men using a similar method. Ten men in acute urinary retention were catheterized and their bladders drained. A known amount of normal saline was then injected and a 10-mL sample obtained from the bladder. The same experiment was conducted in five men using chronic indwelling catheters. A PVR of 98 mL (range 40 to 385 mL) was noted in the 10 men catheterized for acute retention, and a PVR of 15 mL (range 4 to 37 mL) in the 5 men with indwelling catheters.

Based on these findings, it appears that the "gold standard" for measuring PVR, catheterization, is often inaccurate if the person performing the catheterization does not practice techniques to ensure complete emptying. Even then, urine can remain in bladder diverticula and saccules or reflux into the upper tract, causing inaccurate measurements.

Transabdominal ultrasonography, using either a regular B-mode, real-time device or a dedicated unit, is the least invasive method for determining residual urine in men with prostatism. A large number of studies report the accuracy of real-time transabdominal ultrasonography in estimating PVR. All techniques and formulas described in the literature are based on imaging the bladder transabdominally in a transverse and longitudinal orientation, and using either three diameters (length, height, width) or the surface area in the transverse image and the length obtained from the longitudinal image. Bladder volume is estimated by means of the volume formula for a spherical or ellipsoid body.

Although ultrasound is sufficiently accurate for most clinical applications, its cost is a factor. Catheterization is still an appropriate option given the cost and limited office availability of ultrasound units.

However, in addition to standard diagnostic ultrasound instruments for abdominal scanning, there is a much smaller, portable, and less expensive device to measure PVR (BladderScan; Diagnostic Ultrasound). Its reported accuracy is comparable to more expensive ultrasound units and catheterization. With this device, the mean difference between estimated PVR and "true" PVR (that is, by catheterization) was 6.9 mL in 39 measurements taken in 20 children with neurogenic bladders (Massagli, Jaffe, and Cardenas, 1990). In 164 measurements in adult patients, the correlation coefficient was r = 0.79 (Ireton, Krieger, Cardenas, et al., 1990).

Reproducibility

Birch, Hurst, and Doyle (1988) reported that of 30 men with BPH, 66 percent had wide variations in PVR when three measurements were done on the same day using five different formulas to calculate the volume. In 34 percent of patients, there was no difference among the three measurements. In 58 percent, at least two volumes were significantly different. In 8 percent of patients, all three were different. In most patients, two measurements were statistically similar while the third one yielded quite different results. This study proves that the intraindividual variation of the amount of residual urine is greater than the differences between the various formulas utilized.

Bruskewitz, Iversen, and Madsen (1982) found similarly wide variations of the measured amount when they performed repetitive measurements of PVR (repeated 2-5 times) by in-and-out catheterization on 47 men prior to prostatectomy. They also found no correlation between the amount of residual urine and any urethrocystoscopic or urodynamic findings, symptoms, or the presence or absence of a history of urinary tract infections.

The intraindividual variation in PVR measurement is significant, regardless of the techniques used. Although repeated measurements may minimize the error, this is either costly (noninvasive techniques) or uncomfortable (invasive techniques) for the patient.

PVR in the Evaluation of Patients with BPH

Shoukry, Susset, Elhilali, et al. (1975) found that 22 percent of 44 men with significant levels of residual urine on intravenous urography (IVU) films had a "normal" Q[max] (peak flowrate). These data indicate a poor correlation between the volume of residual urine and other urodynamic assessments of obstruction severity.

Griffiths and Castro (1970) measured PVR by radiographic methods in 45 men with prostatism and found a reasonable correlation with the amount determined by catheterization, but no correlation with other clinical data such as prostatic size, Q[max], maximal bladder pressure, urethral resistance, and symptom scores. Most investigators believe that large amounts of PVR urine indicate severe obstruction. Yet, Griffiths and Castro (1970) found that 24 percent of the patients with severe obstruction had less than 50 mL of PVR urine.

PVR was found to be of little help in assessing 117 patients with BPH in a study by Abrams and Griffiths (1979), who considered PVR urine an indication of abnormal bladder function or detrusor decompensation rather than evidence for urethral obstruction.

Neal, Styles, Powell, et al. (1987) found a significant association in 253 men between PVR, age, "below normal" (undefined) Q[max], and high urethral resistance. Weak voiding pressure, however, did not correlate well with PVR. The authors concluded that outflow obstruction is related to the development of increasing amounts of PVR urine.

Predictive Value of PVR

Traditionally, urologists have assumed that increasing amounts of PVR denote BPH progression and are thus an "indication" for surgery. This concept underlies the common inclusion of PVR values in individual State PRO criteria. Unfortunately, data are lacking to support the predictive value of PVR.

Andersen (1982) studied 104 men with BPH and reported two patterns of BPH progression. The slow course was characterized by the development of high levels of PVR that resulted in decompensation of the detrusor muscle and eventually led to urinary retention. The fast course was associated with uninhibited detrusor contractions (UDCs). The amount of PVR, the presence of UDCs, and symptoms correlated poorly in the study. Nevertheless, Andersen recommended PVR as a safety parameter, when measured longitudinally throughout the clinical course of a patient with prostatism.

Jensen, Jorgensen, and Mogensen (1988c) examined the prognostic value of PVR and 14 other clinical and urodynamic variables in relation to outcome of surgery in 120 men with prostatism. They found PVR the second best predictor of outcome, after pressure-flow studies. However, the combination of these two predictors did not allow the authors to predict correctly the outcome in any of 14 patients who failed treatment. The authors did correctly predict all successfully treated patients.

The panel reviewed unpublished data from a randomized trial comparing TURP with watchful waiting (Veterans Affairs Cooperative Study Group). These data demonstrate that PVR does not predict the outcome of surgery, and there appears to be little evidence to support criteria that require a certain amount of PVR before surgery is justified (Wasson, Reda, Bruskewitz, et al., forthcoming ). High PVR did predict, in this trial, a slightly higher failure rate for watchful waiting. However, the majority of men with large residual urine volume did not require surgery during the 3-year duration of the trial.

Conclusions

The volume of PVR correlates poorly with other signs and symptoms of prostatism. The common assumption, that the amount of residual urine predicts which men will or will not improve after surgery for prostatism, is not adequately supported by enough evidence to mandate use of the test. The test is, therefore, optional in the evaluation of men presenting with prostatism.

The panel believes that present review criteria requirements, for certain amounts of residual urine before a prostatectomy is considered appropriate, cannot be justified on the basis of current data. Requiring the presence of residual urine before offering treatment implies that at least partial bladder decompensation must occur before intervention is appropriate.

Nevertheless, although poorly studied in the literature, monitoring of PVR is a reasonable option for men being followed with BPH or for those selecting nonsurgical therapies. It is likely that high residual urines may lead to a slightly higher failure rate for watchful waiting, but repeat measurements should be considered before making treatment decisions based on PVR values.

Prospective evaluation of the predictive value of PVR is needed. Specifically, the risks associated with given volumes must be determined in order to ascertain what PVR values are associated with a risk for urinary retention, infection, or poor treatment outcome.

Methods of Measurement

Pressure-flow tests measure pressure in the bladder during voiding. To monitor intravesical pressure (P[ves]), it is necessary to introduce a recording device into the bladder either transurethrally or suprapubically. A rectal recording device is often used to subtract the intrarectal (that is, intra-abdominal) pressure from the intravesical pressure. The difference between the intravesical and the intrarectal pressure is the detrusor pressure (P[det]), reflecting the pressure generated by the bladder muscle alone.

A detrusor pressure of greater than 100 cm H[sub2]0 at the time of the peak flowrate would indicate obstruction according to Smith (1968). Lower "obstructive" pressures have been recommended by other investigators. Flowrate (Q[max]) and voiding pressure have also been related in the form of a urethral resistance factor, according to a concept derived from rigid-tube hydrodynamics theory (Griffiths, 1975). The urethral resistance factor is calculated as the ratio between the intravesical pressure at Q[max] over Q[max]^2 (R = P/Q^2 or P[det]Q[max]/Q[max]^2). A urethral resistance greater than 0.6 is believed to indicate obstruction.

However, the urethra behaves like an elastic, not a rigid, tube. This limits the value of resistance measurements calculated in the usual fashion (Spangberg, Terio, Ask, et al., 1991). Rather than combining the data in a single factor, the International Continence Society recommends plotting them in the form of a pressure-flow diagram (Abrams, Blaivas, Stanton, et al., 1988). A subgroup of patients with equivocal results may or may not have obstruction, and their poor flow may be due to a nonrepresentative micturition or a weak detrusor. The presence of this equivocal group demonstrates that a single "urethral resistance factor" cannot capture the complex relationship between intravesical pressure and flow.

The value of pressure-flow plots is accepted by many urodynamic experts. Yet there is little standardization in interpretation of these plots and few clear "cutoff" values for defining obstruction as opposed to nonobstruction. Investigators have proposed various ways to present the same sets of data and claim superior differentiation between patient groups (Abrams, Farrar, Turner-Warwick, et al., 1979 ; Jensen, Jorgensen, and Mogensen, 1988b, 1988d ; Schafer, Noppeney, Rubben, et al., 1988 ; Schafer, Rubben, Noppeney, et al., 1989 ; Spangberg, Terio, Ask, et al., 1991). This variability in data presentation and definition has made it difficult to analyze the evidence that supports the use of such pressure-flow plots.

Predictive Value of Pressure-Flow Studies

The most important issue to address is predictive value. To what extent can invasive pressure-flow studies identify patients who will not improve after surgical treatment of their urinary symptoms? To what extent does the addition of pressure-flow studies enable the clinician to separate the truly obstructed from the unobstructed patients? To what extent does this distinction predict the outcome of BPH therapy? Unfortunately, only a limited number of reports present outcome data addressing the value of pressure-flow studies in reducing the symptomatic failure rate of surgery.

Bruskewitz, Jensen, Iversen, et al. (1983) studied prospectively the outcome in 46 patients with prostatism. Patients were divided into two groups with urethral resistance either less than or greater than 0.6. The urethral resistance was correlated with symptom scores and with Q[max], but not with the size or resected weight of the prostate. Eighty-nine percent of the patients with a resistance factor less than 0.6 reported global subjective improvement, compared with 92 percent of patients with a resistance factor greater than 0.6 (not significant). The authors concluded that in patients with moderate prostatism, a cutoff value of R = 0.6 does not predict either indirect health outcomes (Q[max] , detrusor pressure, resistance) or direct health outcomes (symptom score assessment, global subjective symptom assessment) after surgery.

Abrams, Farrar, Turner-Warwick, et al. (1979) examined 190 men (age 47-85) who underwent prostatectomy. Followup data after surgery were available for 152 patients. The decision to perform surgery was based on urodynamic testing, but the report did not clarify which criteria were applied to the decision to operate and whether the surgeons were blinded concerning the preoperative urodynamic evaluation. The patients were considered improved after surgery if their Q[max] was greater than 15 mL/sec at a pressure less than 100 cm H[sub2]0. Subjectively, 88 percent of the patients improved, 12 percent showed no change, and none worsened postoperatively. Objective assessment revealed the following outcomes:

• Improved symptoms, relief of obstruction: 76.0 percent.

• Improved symptoms, partial relief of obstruction: 10.0 percent.

• Improved symptoms, no relief of obstruction: 2.5 percent.

• Symptoms not improved, relief of obstruction: 7.5 percent.

• Symptoms not improved, no relief of obstruction: 4.0 percent.

Objectively assessed, 86 percent of patients improved as compared with the 88 percent who improved subjectively. Q[max] increased from 6.9 [pm]4.1 mL/sec to 20+/-7.1 mL/sec. In an earlier study (Abrams, 1977), the inclusion of urodynamic data in the preoperative evaluation and indication for surgery reduced the subjective failure rate to 12 percent, down from 28 percent when patients were certified as candidates for surgery without the urodynamic data. However, a 28-percent failure rate is significantly higher than that reported in other TURP series.

Another study (Abrams, and Griffiths, 1979) used pressure-flow plots in addition to flowrate measurement. The study found that in about half the cases the patients with prostatism could be correctly classified as obstructed or nonobstructed by Q[max] alone, but that the addition of the detrusor pressure at Q[max] allowed correct classification in two-thirds of patients. Combining the two parameters into a single urethral resistance factor did not help in the two-thirds group. The remaining one-third of the patients were assessed by pressure-flow plot. In many of these patients, both pressure and Q[max] were low, indicating a decompensating detrusor muscle as the source for the low Q[max].

Table 9. Urodynamic variables used to predict outcome (more...)

Table 9. Urodynamic variables used to predict outcome after surgery

Variables	Overall percent correctly classified	Percent unsuccessful patients correctly classified	Percent successful patients correctly classified
Pressure-flow plot (P/Q) and residual urine (PVR)	88.3% (106/120)	0% (0/14)	100% (106/106)
P/Q, PVR, Qmax, detrusor pressure, irritative symptom score	88.3% (106/120)	0% (0/14)	100% (106/106)

Note: Qmax = peak flowrate.

Source: Jensen, Jorgensen, and Mogensen, 1988d.

In a study by Kuo and Tsai (1988), 50 patients (age 51-84, mean age 65) were evaluated urodynamically and by symptom analysis before and after prostatectomy. Symptom analysis was assessed subjectively without a scoring system and postoperative results were reported as "good," "fair," or "worse." The results of the urodynamic tests were not made known to the surgeon.

Table 10. Correlation between preoperative urodynamic (more...)

Table 10. Correlation between preoperative urodynamic data and outcome following surgery

	Outcomes
Patients	Good	Fair	Worse
HPOB	31/33 (94%)	2/33 (6%)
LPOB	2/9 (22%)	6/9 (67%)	1/9 (11%)
LPNOB	1/8 (12.5%)	2/8 (25%)	5/8 (62.5%)
Qmax < 10 mL/sec	18/23 (78%)	4/23 (17%)	1/23 (5%)
Qmax 10-15 mL/sec	14/17 (82%)	1/17 (6%)	2/17 (12%)
Qmax > 15 mL/sec	2/10 (20%)	5/10 (50%)	3/10 (30%)
Overall outcome results	68%	20%	12%

Note: HPOB = high-pressure obstructed bladder, LPOB = low-pressure obstructed bladder, LPNOB = low-pressure nonobstructed bladder, Qmax = peak flowrate.

Source: Kuo and Tsai, 1988.

Schafer, Noppeney, Rubben, et al. (1988) and Schafer, Rubben, Noppeney, et al. (1989) used sophisticated, computerized analysis of pressure-flow data to define passive and dynamic urethral resistance measures. They documented that approximately 25 percent of patients undergoing TURP are not obstructed, based on the investigators' criteria, although these patients may have a low Q[max]. The improvement rate based on objective measures (Q[max] and PVR) was 100 percent in patients categorized as severely obstructed, but was lower in the mildly obstructed and nonobstructed patients. Direct health outcome data were not reported by these investigators.

The Schafer model has been used and expanded by other investigators (Spangberg, Terio, Ask, et al., 1991), but correlation of the objective results with direct health outcomes (subjective results) is again lacking. Others have employed computer programs to analyze pressure-flow plots, but also do not report outcome data (Rollema, van Mastrigt, and Janknegt, 1991 ; Rollema, and van Mastrigt, 1991). The details of the analysis and the computer program are not discussed in an outcomes context. These computer-assisted urodynamic assessments need further evaluation before they can be recommended routinely.

Conclusions

Pressure-flow studies provide much more specific insight into detrusor function and the etiology of voiding dysfunction than do flowrate measurements. However, the limited number of outcome-based investigations performed demonstrate a modest additional value of pressure-flow studies over symptom and flowrate evaluation. The benefit of pressure-flow studies is clearest in those patients who have a Q[max] >15 mL/sec or in whom the initial evaluation suggests bladder dysfunction rather than BPH as the cause of patient symptoms. Moreover, intraindividual and interobserver variability have not been adequately defined for these studies. Although pressure-flow plots are generally recognized to be more informative than measurement of urethral resistance, interpretation of such plots has not been standardized.

Given the uncertainties that exist concerning the pressure-flow study, its use should not be mandated in routine cases at the present time. It is, however, an appropriate optional test in patients for whom the distinction between urethral obstruction and impaired detrusor would affect therapeutic decisions. In addition, pressure-flow studies are appropriate in patients with underlying neurologic disease that may affect detrusor or sphincteric function.

Prospective trials to demonstrate the predictive value of pressure-flow studies are needed before further recommendations can be made.

Test Performance Methods

The International Continence Society (ICS) has established performance criteria for filling cystometry. In general, sterile fluid should be instilled into the bladder with continuous measurement of the intravesical pressure. The bladder should be filled at a moderate rate between 10 and 100 mL per minute. The fluid should be warmed to body temperature except when performing a provocation test (ice water test) with cold fluid.

The patient's response is monitored during filling, especially for urinary urgency sensations such as first sensation of filling, strong sensation, urge to urinate, or uncontrollable urge to urinate. The test is stopped when the patient has a strong urge to urinate. Usually it is performed with the patient in a supine and relaxed position, but he may also be in an upright or sitting position. If there is uncertainty about the presence or absence of UDCs, the examination might be repeated as a provocation test by fast-filling, filling with cold fluid, or filling with the patient in an upright position.

To obtain reliable results, a fluid-inflatable balloon is inserted into the patient's rectum during cystometry to record the intrarectal pressure as a proxy for the intra-abdominal pressure. The difference between the intravesical pressure (P[ves]) and the intra-abdominal pressure (P[abd]) represents the true detrusor pressure (P[det]) measured in centimeters H[sub2]O (P[ves] - P[abd] = P[det]).

Definitions for normal cystometric values have been published in four reports of the ICS (Abrams, Blaivas, Stanton, et al., 1988). Involuntary pressure increases that are greater than 15 cm H[sub2]O above baseline are considered UDCs.

Correlation of UDCs and Irritative Voiding Symptoms

Irritative voiding symptoms, common in men with BPH, are correlated with the presence of UDCs. The high incidence of UDCs in patients evaluated for BPH is well documented. The probability of preoperative UDCs does not differ much among the studies. Analysis of the published data reveals a mean probability of 60.2 percent of preoperative UDCs. This finding is associated with a 90-percent CI ranging from 52 to 68 percent.

From data in three studies reporting the postoperative prevalence of UDCs (Abrams, Farrar, Turner-Warwick, et al., 1979 ; Jensen, Bruskewitz, Iversen, et al., 1984 ; Kadow, Feneley, and Abrams, 1988), the overall mean probability of having UDCs after prostate surgery was calculated at 26 percent with a 90-percent CI ranging from 17 to 39 percent. The data indicate that surgery significantly reduces the prevalence of the problem.

The persistence of postoperative UDCs correlates with persistent symptomatology, and thus treatment failure (McLoughlin, Gill, Abel, et al., 1990). However, the finding of preoperative UDCs is not predictive of persistent posttreatment UDCs.

In practice settings where pressure-flow studies are not available, filling cystometry may give some information on bladder function in patients suspected of having a primary or neurogenic bladder disease. However, the value of the test is inferior to pressure-flow urodynamic evaluation.

Conclusions

UDCs are more common in men evaluated for prostatism than in randomly selected, age-matched controls. This urodynamic finding correlates significantly with the presence of irritative symptoms. Filling cystometry does not add significant new information to the evaluation of men with prostatism. UDCs resolve in most patients after surgery. Only about one-fourth of patients who have UDCs before treatment retain them afterward. Patients whose symptoms do not improve following surgery are more likely to have persistent UDCs.

Although filling cystometry may demonstrate a poorly contractile detrusor in men with primary bladder dysfunction, pressure-flow studies provide much more insight into the interaction between bladder contraction and urethral resistance. Filling cystometry may be considered for men in urinary retention who cannot urinate for a pressure-flow study.

Instrumentation

Urethrocystoscopy can be performed with rigid or flexible instruments. Standard rigid instruments provide clearer visualization, a better field of view, and a larger working channel. As an alternative to rigid instruments, flexible cystoscopes have been developed. Flexible cystoscopy has distinct advantages. The instruments are smaller in diameter and therefore more convenient and more comfortable for the examination of male patients. Flexible urethrocystoscopy can be performed easily as an outpatient procedure in the physician's office.

Correlation with Other Parameters

Simonsen, Moller-Madsen, Dorflinger, et al. (1987) prospectively evaluated 199 BPH patients who underwent prostatectomy. A statistically significant correlation (p <0.05) was found between objective findings (including length of the prostatic urethra, and estimated weight of the prostate gland) and the irritative symptoms as indicated by the patient. Cystoscopic appearance of the prostate and bladder, however, did not correlate with obstructive symptoms.

Andersen and Nordling (1980) investigated the correlation between endoscopic and urodynamic findings in a group of 93 patients with prostatism. The investigators found that prostatic weight (estimated by rectal examination, and by urethrocystoscopy) and the bladder neck to verumontanum distance correlated significantly with the opening pressure, and that the cystoscopically estimated weight and the bladder neck to verumontanum distance also correlated with the maximum flowrate. However, the cystoscopic findings were not related to PVR or detrusor instability. More important, the presence of "cystoscopic obstruction" (occlusion) did not correlate with Q[max] or PVR.

Bladder Trabeculation, Cellule, and Diverticula Formation

Urethrocystoscopy can identify the presence of lower urinary tract complications that may be due to BPH (for example, bladder diverticula) or may mimic the symptoms (for example, urethral stricture disease). The low probability of these findings in men without hematuria, urinary tract infection, or a history of risk factors makes it difficult to justify the procedure for routine evaluation of all men with prostatism.

If the patient has a finding indicative of possible bladder stones (for example, hematuria or recurrent infection), then urethrocystoscopy is indicated. A history of urethritis or prior urethral injury increases the risk of stricture disease and also makes urethrocystoscopy an appropriate consideration.

Remarkably, few studies address correlations between bladder trabeculation symptoms, urodynamics, and the outcome of surgery. Bladder trabeculation may predict a slightly higher probability of failing a watchful waiting management strategy. No report in the literature correlates cellule or diverticula formation with symptomatology or outcome after surgery.

The detection of bladder diverticula may be of therapeutic relevance. For example, the presence of a large bladder diverticulum may influence the type of surgical intervention (open as opposed to closed prostatectomy). However, no data are available to document the sensitivity or specificity of cystoscopy, cystography, intravenous urography, or transabdominal ultrasonography in the evaluation of patients for such diverticula. Moreover, data on the natural history of untreated as opposed to treated diverticula are not available.

Predictive Value

The endoscopic appearance of the bladder and prostate is often felt to be helpful in the decision to treat. Although the linkage between the endoscopic appearance of the lower urinary tract and treatment outcome is poorly documented in the literature, available information suggests that the relationship is minimal. Bladder trabeculation may predict a slightly higher failure rate in patients managed by watchful waiting but does not predict the success or failure of surgery. Urethrocystoscopy may, nevertheless, be useful in determining the technical feasibility of specific invasive therapies. For example, if urethrocystoscopy reveals a large middle lobe, balloon dilation and transurethral incision of the prostate (TUIP) are unlikely to be successful. The decision to perform an open prostatectomy may be appropriately influenced by the shape of the gland, as well as its size. In all of these cases, however, the patient and his physician have already selected invasive therapy. Urethrocystoscopy is therefore performed to select (or rule out) specific techniques, not to determine the need for treatment.

Conclusions

The remote probability of identifying, by urethrocystoscopy, lower urinary tract complications possibly due to BPH, in men without hematuria, urinary tract infection, or a history of risk factors, makes the routine use of this procedure for all men with prostatism questionable. Available data suggest there is a minimal relationship between the endoscopic appearance of the lower urinary tract and treatment outcome. However, urethrocystoscopy may be useful in determining the technical feasibility of specific invasive therapies.

Background

Imaging of the urinary tract prior to prostate surgery has become an integral part of the urologic workup of patients with benign prostatic obstruction.

The frequently made decision to image the urinary tract prior to prostatectomy has been questioned by several authors (Butler, Donnelly, and Komaranchat, 1978 ; Bauer, Garrison, and McRoberts, 1980 ; Christofferson, and Moller, 1981 ; Wasserman, Lapointe, Eckmann, et al., 1987). They each base their recommendation to eliminate imaging, as a routine policy, on their analyses of series of patients with BPH.

Other authors recommend an alternative imaging modality to IVU, namely ultrasonography, to evaluate the urinary tract (Lilienfeld, Berman, Khedkar, et al., 1985 ; Cascione, Bartone, and Hussain, 1987 ; Solomon, and Van Niekerk, 1988 ; Stavropoulos, Christodoulou, Chamilos, et al., 1988). Their recommendation is based on the superiority of sonography in assessing renal lesions seen on intravenous urography and on its harmlessness and noninvasiveness.

Nevertheless, other investigators continue to maintain that routine IVU imaging should remain part of the perioperative workup of BPH patients, and they present data on screened patient populations to support their views (Donker, and Kakiailatu, 1978 ; Pinck, Corrigan, and Jasper, 1980 ; Muzafer, 1986).

Summary of Health Outcomes

One often stated benefit of evaluating the urinary tract prior to BPH treatment is the opportunity to locate lesions in the kidney, such as stones or tumors, that would otherwise have gone undetected. Additional benefits include identification of hydronephrosis, large residual urine, or diverticula that may affect management.

Another stated benefit of a pretreatment imaging policy may be reduction in unnecessary or untimely prostate surgeries. If prostate size is accurately assessed, the risk of undergoing a type of surgery not indicated for a given gland size (TURP as opposed to open prostatectomy) might be reduced. This could affect duration of hospital stay, time of recovery, and cost of care.

Harms associated with a pretreatment evaluation policy include the cost of the procedure, the time spent by the patient away from work, biologic effects (such as radiation effects), risk of allergic or other nonlethal adverse effects due to the contrast medium, and the slight risk of dying from an adverse reaction to the contrast material.

Contrast Media Risk

Table 11. Adverse effects due to intravenous administration (more...)

Table 11. Adverse effects due to intravenous administration of contrast

Source of study	Type of contrast medium	N1	Minor adverse effects[2]	Major adverse effects[3]	Any adverse effects[4]	Death[5]
J Radiol Electrol 1966;47:346	Ionic	912,300			0.00164%
Invest Radiol 1970;5:374	Ionic	318,500		5.000%	0.00251%
Am J Roentgenol 1973;119:832	Ionic	33,000			0.00303%
Am J Roentgenol 1975;124:145	Ionic	81,278			0.00738%
Am J Roentgenol 1982;139:919	Ionic	300,000			0.00133%
N Engl J Med 1987;317:845	Ionic	6,763		0.399%	7.497%	0.00000%
Am J Roentgenol 1989;152:939[6]	Ionic	6,006	2.531%	1.582%	4.113%	0.00000%
Radiology 1990;175:621[7]	Ionic	169,284		0.254%	12.658%	0.00059%
Am J Roentgenol 1989;152:939[8]	Nonionic	7,170	0.586%	0.112%	0.697%	0.00000%
Radiology 1990;175:621[9]	Nonionic	168,363	0.045%	3.134%	0.00059%
Australas Radiol 1988;32:426[9]	Nonionic	30,268	0.02%		0.00000%
Total ionic contrast media:		1,827,131	2.531%	0.303%	7.613%	0.00192%
Total nonionic contrast media:		205,801	0.586%	0.044%	3.034%	0.00048%
Number of patients in study. Minor adverse effects noted (if specified as minor). Major adverse effects noted (if specified as major). Any adverse effects noted (if adverse effects do not fallin specified category). Death reported in connection with administration of contrast medium. Prospective randomized trial. Prospective randomized trial from the Japanese Committee on the Safety of Contrast Media. Causal relationship to administration of contrast material not established. Only patients receiving low osmolar contrast media considered.

Those studies describing the use of high osmolar (ionic) contrast media (HOCM) are generally older than those using low osmolar (nonionic) contrast media (LOCM). In general, the weighted incidences for both minor (mild or minimal) and major (moderate to severe) adverse effects are 5-10 times lower for LOCM than for HOCM. The incidence of "Any adverse effects," as defined in Table 11, is about twice as high for HOCM as for LOCM. The death rate following the intravenous administration of HOCM is about 2 in 100,000 (0.0019 percent), compared with 0.48 in 100,000 (0.00048 percent) following LOCM. However, the mortality rates following HOCM are based in part on a study from 1966 in which 15 of 912,300 patients died and on another report from 1982 in which 4 of 300,000 patients died. All of the patients who died had histories of hypersensitivity to other drugs.

Two randomized trials are available to directly compare adverse effects resulting from intravenous administration of HOCM or LOCM in similar populations: (1) Wolf, Arenson, and Cross (1989) and (2) Katayama, Yamaguchi, Kozuka, et al. (1990), the second by the Japanese Committee on the Safety of Contrast Media (JCSCM).

In the JCSCM report, the relative risk of experiencing major adverse effects for patients receiving HOCM was 5.6, with a 95-percent CI ranging from 4.4 to 7.2 percent. In the Wolf study, this relative risk was 14.2 (95-percent CI 6.9-29.1 percent). A combined analysis of both reports yielded a relative risk of 6.2 (95-percent CI 4.9-7.8 percent). In the JCSCM report, the odds ratio for severe and very severe adverse effects was 0.19 and 0.10, respectively, in favor of LOCM. In the Wolf report, the difference between the incidences was significant at a level of p <0.005 for both moderate and severe adverse effects in favor of LOCM.

Kinnison, Powe, and Steinberg (1989) reviewed 100 randomized controlled trials conducted in humans to compare the safety and efficacy of LOCM as opposed to HOCM. In the 43 trials judged most superior, imaging with LOCM was equal to or of better quality than imaging with HOCM for all applications. No differences were seen in regard to mild adverse effects such as nausea, vomiting, and urticaria. More important, greater cardiovascular changes occurred with the use of HOCM, while nephrotoxicity was the same for both types of media. This is consistent with data from Schwab, Hlatky, Pieper, et al. (1989), who found no difference in nephrotoxicity between LOCM and HOCM in 443 patients undergoing cardiac catheterization.

Despite clear evidence of a better safety profile for LOCM, the universal use of low osmolar contrast media is limited by their substantially higher cost. Powe, Steinberg, Erickson, et al. (1988) revealed that average costs for the hospital of managing mild, moderate, and severe adverse effects were $2.52+/-5.33, $24+/-54, and $910+/-749, respectively. On the other hand, the average difference in material costs ranged from $93 (body computerized tomographic [CT] scan) to $179 (cardiac catheterization), thus making the routine use of LOCM not cost-effective. Patients at risk (over 50 years of age, debilitation, severe cardiovascular disease, severe allergy or asthma, prior reaction to contrast media) should, however, receive the benefit of the lower-risk LOCM (McClennan, 1990).

Prostate Size Estimates

To evaluate the accuracy of various imaging modalities in assessing size and thereby weight of the prostate gland prior to surgery, the panel reviewed a large number of studies using IVU, transabdominal and transrectal ultrasonography, digital rectal examination, and urethrocystoscopy. Findings demonstrate that ultrasonographic (transabdominal, and transrectal) methods allow the most accurate estimation of the prostate size (r = 0.88 and 0.89). If knowledge of the exact prostate size is important prior to a planned procedure, an ultrasonographic evaluation policy is better than an IVU policy (r = 0.195). It is noteworthy that digital rectal examination shows a better correlation with the actual resected weight (r = 0.445) than IVU-based estimation. Urethrocystoscopy is less accurate than either transabdominal or transrectal ultrasonography (r = 0.615).

Upper Tract Findings

Table 12. Incidence of significant findings in series (more...)

Table 12. Incidence of significant findings in series using IVU and US in men with BPH, two series using IVU for other indications, and two series using KUBs in an unscreened population

Designations and types of findings	IVU for BPH[1]	US for BPH[1]	IVU for other indications[1]	KUB in unselected patients[1]
Number of studies	25	9	2	2
Number of patients	6,131	778[2]	10,558	1,000
Mean age	68.4	68.4	N/A	N/A
Normal studies	77.5	68.3	70.4	70.5
Uni- or bilateral hydronephrosis	5.8	6.9	3.8	0.6[3]
Percentage of patients with hydronephrosis who had renal insufficiency	33.3	23.3	N/A	N/A
Poorly functioning kidneys	8.3	N/A	N/A	N/A
Poor image quality	N/A	7.5	N/A	N/A
Renal mass requiring evaluation noted	1.7	N/A	3.1	N/A
Renal cyst diagnosed	4.5	15.0	N/A	N/A
Solid renal tumors diagnosed	0.44	0.51	0.18	N/A
Renal stones	2.1	1.3	11.0	2.8
Ureteral stones	0.5	0	2.1	0.9
Bladder stones	2.7	3.2	0.2	0.1
Bladder diverticula	3.4	3.2	0.3	N/A
Bladder trabeculation	32.1	33.7	0.3	N/A
Ureteropelvic junction obstruction	0.41	1.3	0.45	N/A
Nonfunctioning kidney	0.38	N/A	1.4	N/A
Solitary kidney	0.59	1.89	N/A	N/A
Ectopia	0.77	0.99	0.18	N/A
Horseshoe kidney	0.44	0	0.23	N/A
Chronic pyelonephritis	1.28	2.27	2.43	N/A
Renal scars	1.8	0.5	N/A	N/A
Size difference between kidneys	1.43	1.0	4.8[4]	N/A
Gallstones	1.38	2.19	N/A	2.6
Abdominal aortic aneurysm	0.75	0.88	N/A	1.2

1. Incidence in percent unless otherwise specified.

2. All 778 patients were also examined by IVU and are included in the 6,131 patients.

3. Presumably based on the presence of large renal shadow on KUB.

4. Reflects unilateral renovascular hypertension, 27/558 patients screened with hypertension.

Note: IVU = Intravenous urography, US = Ultrasound, KUB = Kidney-ureter-abdominal radiograph, N/A = Not available.

To place the results obtained in a broader context, four other studies were also analyzed. Two of these studies describe the findings in a large number of patients undergoing IVUs for various reasons other than BPH. Leary, Myers, Greene, et al. (1972) studied 558 hypertensive patients by IVU. Ochsner, Buchtel, and Little (1965) reported the findings in 10,000 patients who underwent IVU within a 5-year period at the Ochsner Clinic. Two reports detail the findings from 500 KUBs (kidney-ureter-bladder abdominal radiographs) obtained on patients admitted for chest diseases (Nishio, Kass, and Levine, 1962) and on a series of 500 patients over 40 years of age (268 male, 232 female) seen as either outpatients or inpatients at a university hospital (Rosenbaum, Lieber, Hanson, et al., 1964).

As shown in Table 12, of all IVUs performed in patients with BPH, 77.5 percent were entirely normal (weighted average). Of all ultrasonography evaluations, 68.3 percent were normal. In the two series of IVUs done for indications other than BPH, 70.4 percent of the imaging studies were interpreted as normal; and 70.5 percent of the KUBs were normal.

Only in 22.5 to 31.7 percent of all patients with BPH were any significant findings identified, which is almost identical to findings in the population that had IVUs done for indications other than BPH (29.6 percent). The percentage of normal imaging studies did not differ in reports in which certain patients were excluded because of conditions such as renal failure, urinary retention, hematuria, and urinary tract infection. In fact, in two reports in which all patients were in urinary retention (Marshall, Singh, and Blandy, 1974 ; Butler, Donnelly, and Komaranchat, 1978), the percentage of normal studies was higher than average (86.6 percent, and 81.3 percent, respectively). Only two reports indicated the number of patients who had some degree of renal insufficiency (Donker, and Kakiailatu, 1978 ; Bauer, Garrison, and McRoberts, 1980). Overall incidence was 12.7 percent (116 of 908).

The incidence of some degree of uni- or bilateral hydronephrosis from these combined studies was 5.8 percent for IVU and 6.9 percent for ultrasonography (weighted averages). In the studies done for indications other than BPH, hydronephrosis was found in only 3.8 percent of patients. It was identified on the KUB in 0.6 percent of patients in the KUB series, presumably by an enlarged kidney shadow. Those series with a 100-percent rate of urinary retention did not have the highest incidence of hydronephrosis (3.0 percent, and 12.3 percent) as compared with all series (range from 0 to 40 percent).

Renal insufficiency in patients with BPH carries a higher surgical morbidity and mortality, but the correlation has not been established for the presence of hydronephrosis without renal insufficiency. Thus, it is important to determine the percentage of patients with hydronephrosis who had renal insufficiency. In the IVU series 33.3 percent, and in the ultrasonography series 23.3 percent, of patients with hydronephrosis had renal insufficiency as defined by the investigators. In these patients, assessment of the presence of hydronephrosis was important because it indicated a higher risk category. However, the two-thirds majority of patients with only hydronephrosis did not measurably benefit from the imaging study.

In 1.7 percent (66 of 3,880) of IVUs, a renal "mass lesion" was suspected, requiring further evaluation by ultrasonography or CT scanning. The incidence of "renal cysts" by IVU was 4.5 percent (232 of 5,111). Ultrasonography detected benign renal cysts in 15.0 percent of BPH cases (117 of 778), and these patients did not require confirmational studies. Assuming that most patients found by IVU to have a "mass lesion" or a "renal cyst" will undergo another confirmational imaging study, approximately 6.2 percent of all patients screened would require such testing at a cost of between $300 (for ultrasonography) and $500 to $800 (for CT scanning).

However, kidneys were poorly visualized by ultrasonography in 7.5 percent of the reported cases. Although this percentage was only mentioned in one study, it can be assumed that a sizable number of patients with suboptimal ultrasonography examinations will require another imaging modality, presumably IVU or CT scanning. These are indirect costs associated with a screening policy employing either IVU or ultrasonography.

The incidence of renal cancer in the United States is approximately 20-50 per 100,000 in the male population between the ages of 60 and 70 (Young, Percy, Asire, et al., 1981). The autopsy prevalence of renal cell carcinoma has been found to range from 0.18 to 0.30 percent (Harvey, 1947 ; Bell, 1950 ; Lucke, and Schlumberger, 1957).

A retrospective study of 558 hypertensive patients (Leary, Myers, Greene, et al., 1972) reports a prevalence of renal cell cancer of 0.18 percent (1 of 558). This study included men and women of different age groups, populations with a lower incidence of renal cancer than that in men between the ages of 60 and 70. Thus, it appears unlikely that the prevalence of renal cell carcinoma in the BPH population is higher than in age- and sex-matched controls. Screening for renal tumors is not justified when based on the assumption of an increased incidence of tumors in patients with BPH.

Bladder Stones

Conclusions

About 70 to 75 percent of all imaging studies obtained in patients with BPH are entirely normal, and only a small fraction of the findings obtained in the remaining 25 to 30 percent actually require active intervention or a change in the approach to the patient. Bundrick and Katz (1986) reported a change in management in 2.2 percent (4 of 180) of patients, based on findings obtained on IVU in a population preselected by excluding men with hematuria, infections, and a history of bladder tumors. Pinck, Corrigan, and Jasper (1980) deferred TURP in favor of a more urgent intervention in 2.5 percent (14 of 557). These data indicate that a change in management would result in about 10 percent of the 25 percent of patients in whom the imaging study is "abnormal."

The presence or history of hematuria, renal insufficiency, urinary tract infection, and/or history of stones or prior urinary tract surgery increases the likelihood that IVU or ultrasonography will demonstrate clinically significant findings (Juul, Torp-Pedersen, and Nielsen, 1989 ; Kreel, Elton, Habershon, et al., 1974 ; Andersen, Jacobsen, and Strandgaard, 1977 ; Christofferson, and Moller, 1981 ; Wasserman, Lapointe, Eckmann, et al., 1987 ; Wilcox, and Mitchell, 1977 ; Butler, Donnelly, and Komaranchat, 1978 ; Morrison, 1980 ; Bauer, Garrison, and McRoberts, 1980).

Donker and Kakiailatu (1978) reported that by screening those men with urinary tract infections, gross hematuria, and renal insufficiency, they would have diagnosed almost all of the abnormal findings in their population of 307 men with BPH.

Although there are no conclusive data on the combined incidence of the important clinical predictors listed above, the panel estimates that approximately one-third of all men with BPH have one or another indication for urinary tract imaging.

Assuming that an indication for renal imaging exists, a number of investigators strongly recommend, instead of IVU, ultrasonography combined with a KUB and a determination of the renal function by measurement of the serum creatinine (Matthews, Quayle, Joseph, et al., 1982 ; Lilienfeld, Berman, Khedkar, et al., 1985 ; Cascione, Bartone, and Hussain, 1987 ; Fidas, Mackinlay, Wild, et al., 1987 ; Hendrikx, Doesburg, Reintjes, et al., 1988 ; Solomon, and Van Niekerk, 1988 ; Stavropoulos, Christodoulou, Chamilos, et al., 1988).

This policy is sound if the level of renal insufficiency prohibits the performance of an IVU (which is rarely the case). Ultrasonography is more specific in determining the nature of a renal mass lesion, thus not requiring as many confirmatory studies as IVU. In a significant percentage of patients, however, the imaging will be inadequate (7.5 percent). For the evaluation of hematuria, IVU has been found to be the more sensitive study.

Indirect Treatment Outcomes

Peak urinary flowrate (Q[max]) and PVR are the indirect outcomes most commonly used to objectively determine efficacy of BPH treatment in clinical research. All active treatments (alpha blockers, finasteride, balloon dilation, TUIP, TURP, and open surgery) increase Q[max] and decrease PVR; both Q[max ]and PVR respond very little to placebo treatment. The degree of improvement in Q[max] and of reduction in PVR is superior with surgical treatment.

The panel's review (in chapter 8 of this Clinical Practice Guideline) of indirect outcomes reported in the literature, specifically of flowrate and PVR, is intended to provide clinicians with useful information for judging the relative efficacies of treatment options. Until better data are available on use of the parameters to predict the occurrence of BPH complications or response to treatment, these measurements should not be used to influence a patient's decision regarding the appropriateness of a given therapy once a diagnosis has been established.

Direct Treatment Outcomes

In considering treatment, a patient may not be especially interested in such indirect outcomes as possible improvement of his peak urinary flowrate or decrease of his residual urine, notwithstanding the importance of these parameters to many physicians. The patient is likely to be more interested in such direct outcomes as his degree of symptom improvement after treatment, the chances of complications or side effects of treatment, duration of any hospital stay, and time lost from work, as well as cost. Results of the panel's review of direct outcomes are reported in chapters 9-17 of this Clinical Practice Guideline.

Patient Preference Analyses

In choosing treatments for a disease such as BPH, which principally affects the quality rather than the quantity of life and where the optimal decision may be dictated by personal values rather than scientific evidence, different patients may have different opinions concerning the benefits and harms of direct outcomes. Depending on how bothered a patient is by his symptoms, one patient may find the risks of surgery or of other treatments acceptable given the potential benefit. Another patient may not find the risk of any therapy acceptable, because he is not bothered by his symptoms or because he is averse to any risk, or both. These differences in how individual patients weigh risks and benefits are clearly shown in studies of patient preferences.

The panel conducted two types of preference analyses, the results of which are detailed in chapter 18. For the first analysis, nonurologic physicians and PhD-level medical researchers over 60 years old served as "proxy" patients. As their initial task, they were asked to rank direct treatment outcomes in order of importance. They ranked the benefit of symptom improvement as the most important outcome. It was followed by the potential harms of incontinence, medical complications of surgery, operative mortality, complications requiring surgery, and impotence.

The proxies were presented scenarios describing patients with mild, moderate, and severe symptoms of BPH as well as a scenario describing a patient in refractory urinary retention. If they envisioned themselves as having only mild symptoms, the vast majority of proxy patients elected no therapy (watchful waiting). In the moderate symptom scenario, proxy patients demonstrated a wide range of preferences. For severe symptoms or urinary retention, a large majority of proxies predicted they would undergo surgical intervention.

A similar pattern of preferences was observed in the second analysis, which surveyed actual patients with a range of BPH symptoms. Most patients with mild symptoms preferred watchful waiting, whereas there was a very wide range of preferences in patients with moderate symptoms. Although surgery was the most commonly elected therapy for those patients with severe symptoms, a significant number elected watchful waiting or another alternative therapy.

It is clear from these studies that the majority of BPH patients with mild symptoms are not sufficiently bothered by their symptoms to accept the risks of therapy, including noninvasive therapies. In patients with moderate and severe symptoms, there is a wide range of preferences probably due to varying levels of "bothersomeness" and risk aversion among individual patients.

Given the variations in patient preferences, and because the physician's values (personal perceptions of benefits, and harms) may be different from the patient's, it is important that the patient's values take precedence in determining choice of treatment. Although patients with severe symptoms are more likely to be bothered, and thus will be more likely to accept the risk of surgery, a considerable number of these patients will still prefer watchful waiting or other therapies.

It is also apparent that the "need" for therapy in patients with symptomatic BPH should, in most cases, be determined by the informed patient rather than by the physician. As made evident above, most patients with mild symptoms do not perceive a need for therapy even with noninvasive approaches. Patients with severe symptoms may choose surgery, but many will have a preference for nonsurgical intervention.

However, an appreciable number of patients with moderate symptoms, after being fully informed, will also elect surgical intervention. Therefore, the concept that patients should only receive surgery if they have failed medical therapy is not appropriate. Some patients with moderate symptoms, and significantly bothered by their symptoms, view the potential benefits of surgery as clearly superior to other forms of therapy and the risks as acceptable.

Analysis of Treatment Options

After an exhaustive review of the benefits and harms for each treatment option, the panel incorporated the results of the patient preference analyses into the guideline development process to help determine the net benefit (or value) of a given treatment modality. The panel concluded that, based on current evidence and its limitations, the following are all appropriate options to offer the moderately or severely symptomatic patient: watchful waiting, alpha blocker therapy, finasteride therapy, balloon dilation, and surgery.

Surgery and Balloon Dilation

Figure 7. Percent improvement in symptom score following (more...)

   Figure 7. Percent improvement in symptom score following different BPH treatments

(Figure 7)

Alpha Blockers

Figure 7

Finasteride

In contrast to most studies, the North American and international phase III finasteride trials (Gormley, Stoner, Bruskewitz, et al., 1992 ; Finasteride Study Group, 1993) obtained the baseline score after a 2-week placebo run-in phase during which the score had already decreased in both the placebo and the drug-treated groups. Furthermore, many patients with low symptom severity were entered into the finasteride studies, thus limiting the ability to show a substantial decrease in the total score. In patients whose symptoms were "mild," it was not possible to demonstrate a substantial decrease in the symptom score from baseline with finasteride.

As with other studies of BPH treatment, the higher the baseline score, the more dramatic the reductions noted in the score in both the finasteride and the placebo-treated arms. The 5-mg finasteride-treated patients with "severe" symptoms in the phase III studies had a decrease in their baseline score by about 6.5 and 7.5 points in the North American and international trials respectively, representing an 18- to 21-percent drop in the total achievable score. The difference between patients treated with the active drug and placebo-treated patients was also larger in the "severe" symptom subgroup.

For finasteride, the maximum effect on the symptom score was achieved between 6 and 12 months and maintained over a 24-month followup in the open extension study (5 mg of finasteride). Overall, 63 percent (North American trials) and 71 percent (international trials) of patients treated with 5 mg of finasteride rated their symptoms "improved," compared with 51 percent and 60 percent of placebo-treated patients, respectively.

In these studies, the total symptom score dropped with finasteride (at 5 mg daily) from 10.1 to 7.5 points (or 28.05 to 20.8, drop = 7.25, linearly transformed to a 100-point scale) in the North American trials, and from 10.6 to 6.7 points (or 29.4 to 18.6, drop = 10.8, on a 100-point scale) in the international trials. The placebo-treated patients in both these studies experienced a drop from 10.1 to 9.1 points (or 28.05 to 25.3, drop = 2.75) in the North American trials, and from 10.6 to 8.0 points (or 29.4 to 22.2, drop = 7.2) in the international trials. These changes represent a reduction in symptom score of 26 percent and 37 percent for the finasteride-treated groups and of 10 percent and 25 percent for the placebo-treated groups (North American, and international studies, respectively).

Alpha Blocker Treatment

The assessment of the incidence of adverse effects occurring on treatment with alpha receptor blocking agents is complicated by what are likely different thresholds used in reporting adverse effects in individual studies. The BPH outcomes data analyzed include two studies using alfuzosin, two studies using indoramin, one study using nicergoline, five studies using prazosin, five studies using terazosin, and two studies using YM 12617. There are seven studies describing the use of the nonselective alpha blocker phenoxybenzamine. Fifteen studies in the data base are placebo controlled (attachment C).

Table 14. Combined analysis of adverse events during (more...)

Table 14. Combined analysis of adverse events during alpha blocker treatment

Treatment modalities	N[1]	Weighted average[2]	Mean[3]	90% CI [4]
Mild adverse effects [5]
Severe adverse effects [9]
Placebo	533	21.8%	11.6%	0.9-40.9%
Phenoxybenzamine (nonselective alpha blocker)	275	31.3%	49.1%	14.9-83.9%
Selective alpha-1 blocker[6]	764	18.85%	13.7%	2.1-38.7%
Alfuzosin (Bayes)[7]	271	34.7%	34.7%	30.1-39.6%
Indoramin (Bayes)[7]	56	8.9%	9.2%	4.2-16.7%
Nicergoline (single profile)[8]	16	0.0%	1.4%	0.01-11.1%
Prazosin (Bayes)[7]	102	16.7%	16.7%	3.2-42.6%
Terazosin (Bayes)[7]	72	29.2%	29.2%	21.2-38.7%
YM 12617	247	2.8%	3.2%	1.02-7.2%
Placebo	580	5.0%	5.0%	3.7-6.7%
Phenoxybenzamine (nonselective alpha blocker)	292	8.6%	9.1%	1.95-23.96%
Selective alpha-1 blocker [6]	908	5.2%	5.2%	4.1-6.5%
Alfuzosin (Bayes)[7]	271	9.96%	9.96%	7.3-13.4%
Indoramin (Bayes)[7]	56	12.5%	12.7%	6.7-21.1%
Nicergoline (single profile)8	16	0.0%	1.4%	0.01-11.15%
Prazosin (Bayes)[7]	102	4.9%	5.05%	2.3-9.4%
Terazosin (Bayes)[7]	178	4.5%	4.6%	2.5-7.6%
YM 12617	285	0.0%	0.08%	0.0007-0.7%

1. Number of patients available for calculation, all treatment arms.

2. Weighted average of outcome based on number of patients with outcome and total number of patients in arm.

3. Mean (50%) or point estimate for probability of outcome to occur.

4. Ninety-percent confidence interval (CI) for probability of outcome to occur.

5. Not leading to discontinuation of treatment.

6. All selective alpha-1 receptor blocker studies combined.

7. Combined by Bayes' formula (if not specified, by hierarchical Bayes' formula).

8. Only one study available for analysis.

9. Leading to discontinuation of treatment.

Due to inconsistencies in reporting and lack of homogeneity in the available outcomes data, there is a considerable difference between the raw and the weighted average for most of the treatment arms under consideration. This assertion holds true for both the mild and severe adverse effects. These factors further support the use of Bayesian methods for combining evidence in which the uncertainty, evident in the wide range of reported incidences of side effects, becomes manifest in a wide CI around the point estimates of these outcomes.

The mean probability of mild adverse effects for all selective alpha-1 blocker studies is 13.7 percent (90-percent CI 2.1-38.7 percent). This estimate is only slightly higher than the mean probability for all placebo arms of 11.6 percent (90-percent CI 0.9-40.9 percent). The probability for mild adverse effects with phenoxybenzamine treatment is significantly higher, with a mean probability of 49.1 percent (90 percent CI 14.9-83.9 percent). The most common mild adverse side effects are mild asthenia (tiredness or weakness), headaches, nasal congestion, and mild dizziness.

The mean probability of severe adverse effects leading to discontinuation of the study ranges from 0.08 percent (YM 12617) to 12.7 percent (indoramin) for selective alpha-1 blockers, with an overall combined mean probability of 5.2 percent (90-percent CI 4.1-6.5 percent). The most common severe adverse effects are orthostatic hypotension, dizziness, and significant asthenia.

Recently published safety data for terazosin are similar to those seen in the combined analysis (Lepor, Auerbach, Puras-Baez, et al., 1992).

This estimate is only marginally higher than the mean probability of 5.0 percent (90-percent CI 3.7-6.7 percent) observed for the 580 patients treated in placebo arms. The mean probability for discontinuation because of severe adverse effects was almost twice as high for the 292 patients treated with phenoxybenzamine (9.1 percent; 90-percent CI 1.95-23.96 percent) than for the selective alpha-1 blockers.

Finasteride

Finasteride was approved by the FDA for treatment of BPH on June 19, 1992. The clinical data discussed in this section are based on five studies conducted by the manufacturer of finasteride (Merck, Sharp & Dohme Research Laboratories, 1992).

About 5 percent (82 of 1,645) of all patients dropped out of the 12-month phase III studies due to lack of therapeutic response or worsening of symptoms. Of patients treated with 5 mg of finasteride per day, 5.2 percent (28 of 543) discontinued the study because of adverse events, primarily sexual adverse events, prostate cancer, and urinary retention. This rate is similar to the 5.0 percent dropout rate in the placebo-treated patients.

Seventeen percent of patients (92 of 543) had an adverse event that may have been drug related according to the investigators. However, only the occurrences of ejaculatory disorders (2.9 percent) and erectile impotence (3.7 percent) were significantly more common in the finasteride-treated patients than in the placebo-treated patients. No deaths were attributable to the use of finasteride, and there is no measurable risk of urinary incontinence or risk of operative complications with its use.

Table 15. Percentages of adverse events in >=0.5% (more...)

Table 15. Percentages of adverse events in >=0.5% of patients (phase III finasteride studies only)

Adverse Event	1 mg finasteride (n = 547)	5 mg finasteride (n = 543)	Placebo (n = 555)
Abdominal pain	0.9	0.9	0.4
Asthenia	0.9	0.9	0.7
Diarrhea	0.7	0.0	0.0
Nausea	0.2	0.4	0.7
Flatulence	0.0	0.6	0.5
Headache	0.5	0.9	0.5
Decreased libido	4.9[1]	3.3	1.6
Dizziness	0.7	0.6	0.7
Skin rash	0.7	0.6	0.2
Ejaculation disorders	2.7[2]	2.9[3]	1.1
Erectile impotence	4.0[1]	3.7[1]	1.1
Orgasm dysfunction	0.2	0.6	0.2
Testicular pain	0.4	0.6	0.4

1. p < 0.01 compared with placebo.

2. p < 0.05 compared with placebo.

3. Decreased ejaculatory volume in 15 of 16 patients.

Concerns have been raised about whether long-term adverse effects might be seen as a result of increases in circulating testosterone (averaging about 10 percent) among finasteride-treated men. These changes in serum testosterone are small and remain within the normal range. Their clinical significance is questionable. Intraprostatic levels of testosterone increase more significantly, but the biologic significance of this finding is uncertain.

The reduction in serum PSA values seen with finasteride treatment as a result of its mode of action have raised concern about whether the early detection of prostate cancer might be impaired among finasteride-treated men. Finasteride decreases serum PSA levels by approximately 50 percent in men with BPH and is known to partially suppress serum PSA in men with prostate cancer (see chapter 3). Yet it is not known whether finasteride treatment actually reduces the information value of PSA among men with BPH or the detection rate for prostate cancer. Furthermore, the benefits of early detection of prostate cancer remain undefined.

Open Prostatectomy

As shown in Table 16, the mean probability for immediate complications following open prostatectomy by any method was calculated at 21 percent (90-percent CI 7.0-42.7 percent). For retropubic open prostatectomy, the mean probability was calculated at 18.7 percent (90-percent CI 5.7-39.8 percent), and for suprapubic open prostatectomy at 25.2 percent (90 percent CI 9.5-47.5 percent). Although the probability for surgical complications appears higher for suprapubic prostatectomy, the difference is not statistically significant.

Table 16. Analysis of immediate surgical complications (more...)

Table 16. Analysis of immediate surgical complications

Treatment modalities	Surgical complications[1]
	N2	Weighted average [3]	Mean [4]	90% CI [5]
	Bleeding intervention [6]
	N2	Weighted average [3]	N [2]	Weighted average [3]
Balloon dilation	323	2.8%	4.8%	1.8-9.9%
Open prostatectomy, any method	9,538	24.5%	21.0%	7.0-42.7%
TUIP	1,243	17.4%	12.1%	2.15-33.3%
TURP	11,693	11.9%	14.95%	5.2-30.7%
Balloon dilation	256	1.95%	323	0.0%
Open prostatectomy, any method	7,761	1.5%	6,989	35.06%
TUIP	1,226	0.5%	1,243	1.2%
TURP	7,184	2.2%	7,603	12.5%

1. Any surgical complications listed in report regardless of severity.

2. Number of patients available for calculation.

3. Weighted average of outcome based on number of patients with outcome and total number of patients in arm.

4. Mean (50%) or point estimate for probability of outcome to occur.

5. Ninety-percent confidence interval (CI) for probability of outcome to occur.

6. Incidence of patients requiring intervention for perioperative

7. Incidence of patients receiving blood transfusion perioperatively (great variation in threshold for transfusion depending on publication date and country of origin).

Note: TUIP = transurethral incision of the prostate. TURP = transurethral resection of the prostate.

Endoscopic Surgery

Investigators have reported that immediate surgical complications occur in 12.1 percent (90-percent CI 2.15-33.3 percent) of TUIP cases.

Because of the wide CIs, there are no statistically significant differences in the probability for complications to occur following treatment with open prostatectomy, TURP, or TUIP. It should be noted that differences in types of complications occurring after any of the individual treatments are considerable; the extent to which this reflects true differences in complications rather than differences in reporting is unclear. Also, perceptions of what constitutes "significant" complications for reporting purposes may differ between investigators and patients.

The data are further confounded by the fact that reports describing the outcomes of open prostatectomy are generally 10 to 15 years older than those describing either TURP or TUIP, and reports describing balloon dilation of the prostatic urethra are all less than 5 years old. Improved surgical technique, improved anesthesia technique and monitoring, and supportive medical care must be taken into account when complications are compared.

Blood Transfusions

As reported in the literature, the probabilities for patients receiving blood transfusions in the immediate perioperative period vary widely. In some studies, the number of patients receiving transfusions seems excessive by current standards. One explanation is the time period of the data reported. Prior to the onset of the acquired immunodeficiency syndrome (AIDS) epidemic, blood transfusions were used more liberally in surgery.

Ranking the open surgical and TURP series by year of publication reveals that those studies published prior to 1980 reported a significantly higher utilization of blood transfusions than those published after 1980. Although not evident in the literature review, it is unlikely that transfusion rates for TURP now exceed 5 to 10 percent. Transfusions have not been reported following balloon dilation.

TURP Syndrome

TURP syndrome, which occurs in approximately 2 percent of patients undergoing TURP (Mebust, Holtgrewe, Cockett, et al., 1989), is characterized by mental confusion, nausea, vomiting, hypertension, bradycardia, and visual disturbances. The syndrome is clearly related to the absorption of the hypotonic fluid used for irrigation during the procedure, with subsequent dilutional hyponatremia (Harrison, Boren, and Robison, 1956). The risk of TURP syndrome is higher if the gland weighs more than 45 grams and resection (operative) time exceeds 90 minutes (Mebust, Holtgrewe, Cockett, et al., 1989). The condition can be treated by diuresis or the infusion of hypertonic saline.

Epididymitis and Urinary Tract Infection

Table 17. Combined analysis of rates of epididymitis (more...)

Table 17. Combined analysis of rates of epididymitis and urinary tract infection during followup

Treatment modalities	Epididymitis
	N[1]	Median[2]	90% CI[3]
	Urinary tract infection
Balloon dilation	127	0.7%	0.02-4.15%
Retropubic prostatectomy	3,583	2.45%	0.8-5.5%
Suprapubic prostatectomy	3,129	3.6%	0.4-12.6%
Open prostatectomy, any method	6,963	2.6%	0.4-8.2%
TUIP	958	3.04%	0.09-16.3%
TURP	12,677	1.0%	0.09-4.5%
Balloon dilation	291	4.2%	0.25-18.2%
Retropubic prostatectomy	688	2.6%	0.1-12.9%
Suprapubic prostatectomy	1,595	12.5%	5.0-24.3%
Open prostatectomy, any method	2,566	13.4%	2.13-31.6%
TUIP	477	12.45%	1.6-38.25%
TURP	9,946	15.5%	3.4-38.3%

1. Number of patients available for calculation.

2. Median (50%) probability for outcome to occur.

3. Ninety-percent confidence interval (CI) for probability of outcome to occur.

Note: TUIP = transurethral incision of the prostate. TURP = transurethral resection of the prostate.

Patient Selection

Reports on treatment failure and retreatment rates vary considerably depending on the patient population initially selected for treatment, thereby introducing the risk of selection bias. After other options were carefully considered, it was decided to report retreatment/failure rates based on the assumption that patients within each treatment group were similar. This decision was made in order to permit utilization and analysis of the existing data.

Accuracy and Completeness of Followup

The duration of followup is often difficult to assess in published reports of treatment outcomes. Because there is no accurate method to correct for incompleteness of followup, the panel elected to take data at face value and not attempt to adjust the results for possible underreporting of failures.

Accuracy of Retreatment Data Reported

The reported data are not always clear regarding patient choices of retreatment after treatment failure and can lead to underestimation or overestimation of retreatment rates. In the case of watchful waiting studies, patients reported as treatment failures could be assumed to undergo surgical treatment for BPH either to relieve symptoms or because of urinary retention. In the case of alpha blocker reports, however, this assumption may not be accurate. According to data reviewed by the panel, all patients who dropped out of alpha blocker and finasteride studies because of lack of efficacy may not have undergone surgery. Some simply stopped taking the drug and adopted watchful waiting programs. Thus, converting treatment failure rates into rates of patients undergoing surgery might overestimate the number of patients going on to surgical treatment for BPH.

Those patients who failed balloon dilation are reported in most studies to have proceeded to TURP or open prostatectomy. Failure data and retreatment data are parallel for this modality. However, the large-scale claims data base studies reporting on retreatment rates following TURP or open prostatectomy might have overestimated to some degree the true retreatment for recurrent or persistent BPH. Up to 3.7 and 4.3 percent of patients following TURP and open prostatectomy are estimated to require surgical intervention for either bladder neck contracture or urethral stricture disease. It was impossible for the panel to accurately assess what fraction of patients in the claims data analysis underwent correction of such surgically induced complications, in contrast to those patients who truly needed retreatment for BPH. The data were accepted at face value with the understanding that the retreatment data, depending on the source, may be biased toward under- or overestimation.

Watchful Waiting

As background to examining probabilities of failure for watchful waiting, it is useful to look at overall probabilities of prostatectomy. Several reports are available. Glynn, Campion, Bouchard, et al. (1985) reported the results of the VA Normative Aging Study, a cohort study of 2,036 volunteers in the Boston area. The probability, for men age 60 to 69 years, of undergoing a prostatectomy within 3.5 years following enrollment was 4.1 percent. The cumulative probability, for a 40-year-old man who lived to age 80, of having surgery for BPH was projected to be 29.2 percent. A prior clinical diagnosis of BPH in these unselected men was an important predictor of surgery (adjusted odds ratio 3.52, 95-percent CI 1.93-6.42 percent).

Sidney, Quesenberry, Sadler, et al. (1991) found, in an analysis of a data base from the Northern California Kaiser Permanente Medical Care Program, that urinary symptoms were predictive of prostatectomy, especially during the first 5 years of followup. The relative risk associated with having two obstructive symptoms, compared with having no obstructive symptoms, was 8.6 in the first 5 years (95-percent CI 6.2-11.9 percent).

Based on data from New Haven, Lytton, Emery, and Harvard (1968) estimated the probability of a 40-year-old man requiring an operation for BPH if he lived to 80 years of age as being approximately 10 percent. The risk for a man age 55 of undergoing prostate surgery before age 80 was 9.7 percent.

Based on data from the Baltimore Longitudinal Study of Aging, Arrighi, Guess, Metter, et al. (1990) calculated incidence rates of surgery for BPH per 1,000 men per year. The incidence rates increased steadily with age and were 14.2 per 1,000 for 60- to 64-year-old men and 36.8 per 1,000 for 80- to 84-year-old men.

These data demonstrate that the risk of a man undergoing prostatectomy increases not only with age, but also with the clinical diagnosis of BPH and the presence of symptoms associated with BPH.

The closest subpopulation in these studies comparable to a group of men with BPH being followed by a watchful waiting strategy is a group of men in the BLSA study who had clinical signs of BPH and enlarged prostates on digital rectal examination. For this group, the probability of subsequent surgery was 36.8 percent. Ball, Feneley, and Abrams (1981) followed, over 5 years, 107 patients with prostatism in whom prostatectomy was not clinically indicated at the time of initial presentation. Of these men, 10 required surgery for worsening of symptoms while 97 remained untreated. Two of these 10 men developed acute urinary retention over the 5-year followup. The retreatment probability in this group of men followed with observation (or watchful waiting) was 8.9 percent at 60 months.

Clarke (1937) followed a total of 93 men for a different length of followup and observed the percentage of patients that required prostatectomy during that time. In group 1 (doubtful diagnosis, n = 31 patients), followed for a mean duration of 5.3 years, the 60-month retreatment probability was 32.1 percent. In group 2 (definite diagnosis of BPH with small residual urine, and no complications, n = 36), the retreatment probability was 30 percent at 2.8 years. In group 3 (definite diagnosis with PVR over 100 mL or other complications, n = 15), the retreatment probability was 48.7 percent at 2.64 years. For group 4 (patients during or following acute or chronic urinary retention, n = 11), the retreatment probability after a followup of 2.64 years was 16.7 percent.

Useful information on watchful waiting comes from a study by Craigen, Hickling, Saunders, et al. (1969). Of 251 patients initially registered by 59 doctors in East Anglia, England, 123 presented with urinary symptoms indicative of prostatism but no retention. An analysis of those 123 patients revealed a retreatment probability (cumulative incidence of failure) of 55.8 percent at the end of the 5-year followup.

The risk of a patient's failing watchful waiting remains poorly defined. In general, the published watchful waiting case series describe the experience of men on surgical waiting lists. In this group of patients with severe symptoms, disease progression may be more rapid than in those men with mild-to-moderate symptoms. Certainly a failure rate of nearly 56 percent, as seen in the Craigen study, is higher than anticipated for men with only mild symptoms.

The Craigen study demonstrates a higher risk of failure during the first 1 to 2 years of followup, which subsequently decreases; that is, the incidence density decreases over time. Assuming similar decreases in the failure rates of BPH medical and surgical treatment over time, projection of long-term failure rates are problematic.

Balloon Dilation

Several studies provide short-term data on balloon dilation failure rates up to 12 months (Shipman, and Akilie, 1967 ; Klein, Perez-Marrero, Bowers, et al., 1990 ; Goldenberg, Perez-Marrero, Lee, et al., 1990 ; Daughtry, Rodan, and Bean, 1990 ; Gill, Machan, Allison, et al., 1989). In a longer term report from Wasserman, Reddy, Zhang, et al. (1990), the followup extended to 3 years (mean followup 16.2 months). Seven of 73 patients were in urinary retention at the time of the procedure. According to the authors, prostatectomy was required at the most recent followup in 17 subjects. The rate of failure (incidence density) for year 1 was 20 of 588 (or 0.034) person-months. In year 2, two more patients required prostatectomy for an incidence density of 0.0065, and in year 3 one more patient required prostatectomy for an incidence density of 0.0086. Combining the data from years 2 and 3 produces an incidence density of 0.0071.

The assumption that the retreatment probability would follow in all other patient populations a similar pattern to that described by Wasserman, Reddy, Zhang, et al. (1990) can be challenged. Clearly, more long-term studies (3-5 years) of balloon dilation are required to accurately define failure and retreatment risk.

Alpha Blockers

Failure and retreatment rates following therapy with alpha blocking agents are difficult to estimate. At the time of the panel's review, data on patients who dropped out of alpha blocker studies because of lack of efficacy were available mainly from two reports (Jardin, Bensadoun, Delauche-Cavallier, et al., 1991 ; Lepor, Henry, and Laddu, 1991). Each of these reports utilizes a specific alpha-1 receptor blocking agent (alfuzosin, and terazosin, respectively). Another source of data on terazosin is a report by Lepor and Knapp-Maloney (1991).

It cannot be determined whether the patients who dropped out because of lack of efficacy did in fact proceed to surgical treatment for BPH or to any other treatment for BPH. For the analysis presented in this section, the panel assumed that patients who failed to symptomatically improve under any given treatment protocol would proceed to another form of BPH therapy.

The report by Jardin, Bensadoun, Delauche-Cavallier, et al. (1991) is a randomized, controlled double-blind study examining the long-term efficacy and safety of alfuzosin in 251 patients treated with active drug, compared with 267 patients treated with placebo over 6 months. During this treatment period, 6.8 percent of patients dropped out of the alfuzosin group because of lack of efficacy, compared with 14.6 percent who dropped out of the placebo arm for the same reason. Of the alfuzosin-treated patients, 0.4 percent, compared with 2.6 percent of placebo-treated patients, went into urinary retention during the treatment period, presumably requiring surgical treatment.

Two terazosin studies are of particular interest in the report by Lepor, Henry, and Laddu (1991), which summarizes several studies. Designated MK-006 and MK-013, the two are comparable in their inclusion and exclusion criteria as well as in their study design (26-week treatment interval). The maximal dose of terazosin used was 10 mg per day. The MK-006 study enrolled 47 men and the MK-013 study 57 men, for a total of 104 men. There were three premature withdrawals in the MK-006 study because of worsening of BPH symptoms and none in the MK-013 study. For the two studies together, there was a 2.9-percent withdrawal rate because of worsening of BPH symptoms.

Extrapolating from 6-month data and assuming a steady dropout of patients because of lack of efficacy over 5 years or 60 months, the cumulative incidence of failure at 5 years would be 39 percent (90-percent CI 13-70 percent). The projected 5-year failure rate drops significantly to 13 percent (90-percent CI 4-31 percent) if the incidence density of failure past 6 months is decreased to resemble the 20 percent of initial failure rate seen in the watchful waiting study by Craigen, Hickling, Saunders, et al. (1969).

In the panel's opinion, the long-term failure rate of alpha blocker therapy is probably intermediate between the above two estimates (13 to 39 percent). Long-term trials are needed to definitively address the question.

Lepor and Knapp-Maloney (1991) reported on 45 men with BPH treated with a maximal dose of 5 mg of terazosin over 24 months. At 24 months, 47 percent remained on therapy. The remainder discontinued the drug for the following reasons: 13 percent for adverse events, 4 percent for recurrence of symptoms, 4 percent not returning for followup, 4 percent noncompliant, 4 percent for other reasons. The largest cohort, 22 percent, discontinued for a failure to maintain symptomatic improvement. (Percentages do not add to 100 percent because of rounding.)

Finasteride

About 5 percent of all patients treated with finasteride dropped out of the 12-month phase III studies because of lack of therapeutic response or worsening of symptoms. Assuming that all the dropouts would seek alternative treatment options, this suggests a 5-percent 1-year failure or retreatment rate.

Fifty-three patients had prostate surgery while participating in all finasteride clinical studies; 18 of the 53 had urinary retention preoperatively. Other reasons for surgery included elective surgery, lack of improvement, worsening condition, no reason specified, and prostate cancer. In the phase III studies, 11 of 1,000 active-drug-treated patients had surgery within 12 months (5 for urinary retention, 6 for other reasons), while 8 of 555 patients on placebo had surgery (2 for urinary retention, and 6 for other reasons).

After the blind was broken in the phase III studies, all patients were given the option to continue on active drug in a open extension of the study. In this open extension, 13 of approximately 300 finasteride-treated patients underwent surgery (5 for urinary retention, 8 for other reasons) presumably within the 12 months after the study was unblinded. Five of these 13 patients had previously been on placebo; the other 8 had been on 1 or 5 mg of finasteride. The majority of the "treatment failures" in the open extension period presumably occurred soon after unblinding. These patients did not likely want to discontinue the blinded portion of the study, but elected to proceed with surgery after the study was unblinded. It would therefore be incorrect to calculate failure rates based on these data.

Five percent of patients dropped out due to lack of efficacy, and an additional 1 percent underwent surgery, for a total failure rate of 6 percent in the first 12 months of treatment. If treatment failure on finasteride occurs at a constant rate over 5 years, the predicted cumulative probability of failure would be 27 percent (90-percent CI 25-29 percent). If, however, the failure rate declines similarly to that seen in other studies (20 percent of initial rate), the projected cumulative failure rate would be 10 percent (90-percent CI 9-12 percent) at 5 years. The rather low "treatment failure" or "retreatment" rate may be partially explained by the fact that about 30 percent of patients in the phase III studies were only mildly symptomatic at entry.

TUIP

Nine reports are available in the panel's BPH outcome data base to assess the treatment failure rate and the need for retreatment following TUIP (Hellstrom, Lukkarinen, and Kontturi, 1986 ; Dorflinger, Oster, Larsen, et al., 1987 ; Larsen, Dorflinger, Gasser, et al., 1987 ; Nielsen, 1988 ; Delaere, Debruyne, and Moonen, 1983 ; Katz, Greenstein, Ratliff, et al., 1990 ; Mobb, and Moisey, 1988 ; Edwards, Bucknall, Pittam, et al., 1985 ; Orandi, 1990).

Orandi (1990) published the data from his 18-year experience with 753 TUIP procedures in a format that permits an accurate calculation of the incidence density or failure rate over a period from 3 to 60 months. By contrast with other long-term BPH treatment studies, the rate did not decline significantly in years 2 through 5. However, the year-1 failure rate was the highest. If the Orandi data are used to project the 5-year failure rate for all the reported TUIP cases, the cumulative rate was 13 percent (90-percent CI 5-26 percent). If only those studies actually reporting 5-year data are analyzed, the cumulative failure rate is 9 percent (90-percent CI 1-28 percent). Confidence intervals for these two analytical approaches are almost identical.

TURP and Open Surgery

Treatment data have been reported for 44,832 patients treated by TURP and 17,065 patients treated by open surgery (Roos, and Ramsey, 1987 ; Roos, Wennberg, Malenka, et al., 1989 ; Taylor, and Krakauer, 1991 ; Singh, Tresidder, and Blandy, 1973 ; Ball, and Smith, 1982 ; Aalkjer, 1965 ; Stephenson, Chute, Guess, et al., 1991 ; Meyhoff, and Nordling, 1986).

The study by Roos and Ramsey (1987) reported retreatment data on 1,855 patients following TURP and 844 patients following open prostatectomy. Over an 8-year followup period, 16.8 percent of the TURP patients and 7 percent of the open prostatectomy patients underwent a second prostatectomy. Roos, Wennberg, Malenka, et al. (1989) reported on 8-year retreatment rates for 54,077 patients following TURP and open prostatectomy in Denmark, the Oxford region of England, and Manitoba, Canada. The probability of undergoing a second prostatectomy ranged from 5.0 percent (Denmark, and Oxford region) to 15.5 percent (Manitoba, Canada). For open prostatectomy, the 8-year retreatment probabilities ranged from 1.8 percent (Oxford region) to 4.5 percent (Denmark).

In the data presented by Taylor and Krakauer (1991), the probability of having a repeat prostatectomy after 2 years of followup was 4.7 percent following a TURP and 1.84 percent following an open prostatectomy. Singh, Tresidder, and Blandy (1973) reported retreatment rates on 935 patients following TURP and 217 patients following open prostatectomy. Within 1 year of the initial operation, 2.8 percent in the TURP group required retreatment compared with 0.5 percent in the open prostatectomy group. Stephenson, Chute, Guess, et al. (1991) studied the outcome of surgery in 330 men undergoing prostatectomy for BPH. The probability of reoperation at 5 years was 12.9 percent, the highest reported rate in any of the studies reviewed.

The panel believes there are ample published data overall to reasonably define the retreatment (failure) rates for TURP and open prostatectomy. The cumulative retreatment probability at 5 years for TURP is 10 percent (90-percent CI 9-11 percent) and for open prostatectomy, 2 percent (90-percent CI 1-4 percent). These retreatment rates undoubtedly included reoperation for complications such as bladder neck contracture (see chapter 13).

Conclusions

Open surgery appears to have the lowest probability of treatment failure and/or need for retreatment within 5 years, when compared with other treatment modalities. The probability for failure requiring retreatment remains constant throughout at least 5 to 8 years of followup for TURP, TUIP, and open prostatectomy. By contrast, balloon dilation and watchful waiting demonstrate a high failure rate the first year and later a decrease in the failure rate throughout the remainder of the followup to 5 years.

Surgical intervention by any technique has a lower retreatment rate than does initial treatment with alpha blockers, finasteride, balloon dilation, or watchful waiting. Probability distributions for these nonsurgical modalities overlap to the extent that there are no statistically significant differences in probabilities of treatment failure. The long-term failure rates for watchful waiting, medical therapy, and balloon dilation remain to be defined in long-term studies. To date, there is no evidence to demonstrate clearly that medical therapy or balloon dilation will lower the future risk of surgery for men being managed by watchful waiting.

Mortality for 67-Year-Old Men

The average age for surgical treatment for BPH is 67 years. At age 67, the risk of death within the next 3 months is 0.8 percent (U.S. Decennial Life Tables, 1990). If all surgical series correctly identified any patient dying during a followup period of 3 months, it would appear that causes not directly related to the treatment of BPH account for about 0.8 percent of the estimated probability of dying within this time period. However, followup is not complete in most surgical series. In addition, there are no data to determine whether men treated for BPH are, on average, healthier (with a lower risk of death) or sicker (with a higher risk of death) than all men represented in the life tables. It is therefore not possible to assume that 0.8 percent mortality can be subtracted from the estimated mean probability of dying within 3 months after surgery, so as to derive the probability of dying directly due to the treatment for BPH. For information purposes in the Balance Sheet (Attachment B), the probability of dying within the next 3 months for a 67-year-old man is listed as 0.8 percent in the columns for alpha blockers and watchful waiting.

Perioperative Mortality: Open Surgery

Early postoperative mortality is reported in 21 studies of treatment outcomes following open surgical removal of the prostate. The total number of patients included is 25,022. Based on the large number of patients available for this analysis, and the finding that reported incidences range only from 0 percent to a maximum of 5 percent, the range of uncertainty is relatively narrow. The risk of early postprostatectomy death is 2.4 percent (90-percent CI 1.0-4.6 percent). The risk appears to be higher for suprapubic than for retropubic enucleation of the prostate, although this finding may not be due to differences between the two operations.

Perioperative Mortality: TURP

Figure 9. Probability of perioperative mortality following (more...)

   Figure 9. Probability of perioperative mortality following TURP

(Figure 9)

The study contributing the most patients to this analysis reports an overall combined probability for early postoperative deaths of 2.91 percent (Roos, Wennberg, Malenka, et al., 1989). All other studies reporting perioperative mortality following TURP result in a probability of 1.18 percent. Because of the large number of patients involved in the study by Roos, Wennberg, Malenka, et al. (1989), the overall combined probability results in a mean estimate of 1.5 percent (90-percent CI 0.5-3.3 percent).

Perioperative Mortality: Balloon Dilation

Figure 10. Probability of perioperative mortality following: (more...)

   Figure 10. Probability of perioperative mortality following: (1) balloon dilation, (2) open surgery—all modalities, (3) TUIP, and (4) TURP

(Figure 10)

Comparison of Treatment Modalities

(Figure 10)

Perioperative Mortality: Conclusions

There are many confounding factors that prevent firm conclusions from being drawn regarding mortality following BPH surgery. Patient selection bias may influence relative mortality rates more than true differences in the risks of TURP or open prostatectomy. For example, statistically, the average age of the patients reported does not differ significantly between open prostatectomy and TURP. Whether the patients treated by open surgical enucleation of the prostate represent a population that is in general less healthy is impossible to determine from this retrospective analysis of the literature. Patients selected for open surgery rather than TURP may be a healthier subpopulation. This would lead to the conclusion that open prostatectomy is associated with a higher early postoperative mortality rate. For TUIP, however, the mortality risk is clearly lower than for the other surgical interventions.

Laser Prostatectomy

Laser energy has been utilized to destroy neoplastic tissue in a variety of organ systems, and neodymium yttrium aluminum garnet (YAG) laser technology has been used successfully to treat BPH. Initial experiences with bare laser fibers were disappointing, primarily because of their inability to penetrate deeply into prostatic tissue. The first major advance was the development of right-angle laser fibers that permit delivery of the energy at right angles to the fiber.

Right-angle YAG fibers can be guided under direct vision through a cystoscope or by transurethral ultrasound imaging. Investigators in the field do not agree on the optimal technique of energy delivery. Some of the laser technologies produce coagulation necrosis with delayed slough of tissue. Other lasers result in immediate tissue vaporization and ablation. The relative benefits and risks of these two approaches have not been adequately defined. Moreover, the relative advantages of visual guidance compared with ultrasound guidance of the laser fiber itself are uncertain.

Current evidence from short-term, noncontrolled trials suggests that laser prostatectomy produces significant symptom score improvement. Flowrate improvements, however, are not equivalent to those produced by TURP. Blood loss is clearly less with laser prostatectomy, and TURP syndrome is totally avoided; but patients tend to require longer periods of catheterization and have more temporary pain on urination than seen following TURP.

The true benefits and risks of laser prostatectomy must be defined in clinical trials that compare the technology directly with TURP. Long-term trials will also be needed to determine retreatment rates and cost-effectiveness.

Hyperthermia and Thermal Therapy

Hyperthermia and thermal therapy of the prostate utilize electromagnetic waves with a frequency greater than 200 MHz (microwaves) to induce heating and presumably tissue damage. Tissue penetration depends on the frequency of the electromagnetic waves. The lower the frequency, the deeper is tissue penetration. Although microwave has been the primary modality utilized to heat prostatic tissue, recently radiofrequency and high-intensity ultrasound have been used for the same purpose.

To destroy normal tissue, a temperature in excess of 45 degreesC is probably required. The administration of temperatures above 45 degreesC is referred to as thermal therapy, and treatment to temperatures below 45 degreesC is referred to as hyperthermia. Temperatures in excess of 45 to 50 degrees C are known in animal models to produce tissue coagulation, but temperatures in the hyperthermia range have no clearly demonstrable prostatic effects. The principle of hyperthermia, that neoplastic cells are more sensitive to slight elevations of temperature than normal cells (which has been validated in cancer model systems) may not be operational in the benign hyperplastic prostate. Thus, the rationale for hyperthermia in the treatment of BPH remains hypothetical. Thermal therapy, by contrast, has the potential to produce actual tissue destruction.

Several clinical trials have been performed with both transurethral and transrectal hyperthermia. The results are mixed, and there is no consensus on whether hyperthermia has an effect in the prostate above that seen with sham treatment. Recognizing the limitations of hyperthermia, most investigators now utilize higher temperatures.

Several industrially sponsored trials on transurethral thermal therapy are ongoing in different countries, including the United States. Although preliminary, these initial studies suggest that transurethral thermal therapy, at least in the short term, produces symptom and flowrate improvement superior to sham (placebo) treatment. However, the magnitude of improvement is at best intermediate between medical therapy and surgery. In addition, a significant number of patients develop urinary retention after the procedure, requiring short-term catheterization. Finally, patients may have urinary discomfort for some period of time after the procedure. Investigators are now altering the treatment times and energy levels for thermal therapy to try to achieve improvements closer to that seen with surgery.

Thermal therapy is a rational treatment approach warranting further investigation. Long-term, controlled trials are required to determine the effectiveness of this approach.

Prostatic Stents

Prostatic stents are metal devices that can be placed into the prostatic urethra, under either endoscopic or fluoroscopic control. The two stents currently under evaluation are made of biologically inert metal. When expanded into the prostatic urethra, they partially relieve the obstruction from the surrounding prostatic tissue. Over a period of a few weeks to a few months, the stents are covered with normal transitional cell epithelium. To date, most of the patients treated with prostatic stents have been in urinary retention. A majority of patients have been able to void successfully after placement of the stent.

There have, however, been problems with encrustation (calcification) of the stents, as well as some degree of perineal pain and discomfort on urination. Clinical trials of prostatic stents are still ongoing. Long-term efficacy and safety remain to be demonstrated. More important, it remains unclear whether prostatic stents have applications in men with symptomatic BPH who have not developed urinary retention.

Androgen Receptor Antagonist Therapy

Currently being evaluated for the treatment of BPH are several drugs that block the action of androgens at the level of the prostatic cell by inhibiting the binding of testosterone and dihydrotestosterone to the androgen receptor in the nucleus of the cell. In preliminary trials, androgen receptor antagonists have demonstrated some degree of prostatic involution, with improvement in both symptoms and urinary flowrates. However, the toxicity of androgen receptor antagonists is significant. Approximately one-half of patients develop breast tenderness or gynecomastia. In addition, one androgen receptor antagonist has a significant incidence of hepatotoxicity. Androgen blockade produces an increase in serum luteinizing hormone levels that leads to further testosterone production in the testes. Some of this increased testosterone is aromatized into estrogen. Testosterone-estrogen ratios in the breast are thus significantly altered. If androgen receptor antagonists could be developed that do not elevate serum gonadotropins, they may be attractive forms of therapy.

5-Alpha Reductase Inhibition

Several new types of 5-alpha reductase inhibitors are undergoing testing. There are presently no data to suggest that any of the new inhibitors will be substantially better or safer than finasteride.

Alpha Blocker Therapy

Several long-term clinical trials are now in progress to assess the efficacy of alpha blocker therapy over placebo. Published trials are limited to 3-6 months of therapy. The results of the longer term trials will be important to determine whether treatment response deteriorates with time, and to determine actual probability of treatment failure and cost-effectiveness.

It is now recognized that the primary adrenergic receptor in the prostate is the alpha 1C subtype. This observation may well catalyze the development of alpha 1C antagonists that may be more selective for prostatic smooth muscle relaxation and that would presumably have fewer peripheral side effects such as asthenia and orthostatic hypotension.

Aromatase Inhibitor Therapy

In the canine model of BPH, estrogens are synergistic with androgens in the production of experimental BPH. This observation has led to the hypothesis that estrogens are important in both the initiation and maintenance of human BPH. Small levels of circulating estrogen in men are derived from conversion of testosterone to estrogen in peripheral tissue (primarily fat). This testosterone-to-estrogen conversion is catalyzed by the enzyme aromatase. Several inhibitors of the aromatase enzyme have been developed and tested in BPH clinical trials. No results have been published to date from these ongoing clinical trials. Aromatase inhibitor therapy is known to produce significant increases in serum testosterone levels. Many investigators are concerned that the beneficial effect of lowering of estrogen within the prostate may be tempered by stimulation of androgen-dependent growth processes.

Combination Therapy

Given the independent mechanisms of action of alpha receptor antagonists and hormone withdrawal therapies, combinations of these two approaches are rational treatments to consider. Currently, the Department of Veterans Affairs is conducting a large, multicenter, randomized clinical trial to determine the efficacy of finasteride plus terazosin, compared with single agents. Until the results of this trial are available for review, combination therapy cannot be recommended for routine practice.

Introduction

The decision diagram on the following pages is provided as a framework for diagnosis and treatment. It is not intended as a rigid pathway that must be followed in all cases. Individual patients will present in whom deviations from these policies are appropriate. In such circumstances, the health care provider should exercise clinical judgment and act in the patient's best interest.

Explanation of Decision Diagram

The terms recommended, optional, and not recommended indicate degrees of desirability for specific diagnostic interventions. The terms standard, guideline, and option indicate intended degrees of flexibility for treatment policies. Patients with prostate enlargement without symptoms of prostatism should not be evaluated for BPH treatment, but reassessed periodically.

1. Initial Evaluation (Recommended)

• Detailed History: Urinary tract symptoms, prior urinary tract surgery/ other treatment. Medical disease, to evaluate fitness for BPH treatments.

• Physical examination: Includes focused neurologic examination as well as digital rectal examination.

• Digital rectal examination (DRE): Prostate size and/or palpable abnormalities. Anal sphincter tone.

• Urinalysis: Dipstick (including test of bacteriuria, hematuria, and/or pyuria) or examination of the spun sediment.

• Creatinine: Elevated value suggests need for upper urinary tract imaging, preferably by ultrasonography.

• Prostate-specific antigen (PSA): Optional.

2. Prostate Surgery (Guideline)

• There is a treatment-policy guideline for surgery if any of these urinary tract conditions are present secondary to BPH.

3. Quantitative Symptom Assessment

• Ask patient to complete AUA Symptom Index questionnaire (Recommended).

• Put completed AUA form in medical record.

4.Mild Symptoms (Symptom Score = 0-7)

• Offer watchful waiting only (Standard).

• Reassure patient.

• Reassess periodically.

• Risks outweigh benefits for all treatments except watchful waiting.

• Most patients prefer watchful waiting.

5. Moderate Symptoms (Symptom Score = 8-19) to Severe Symptoms (Symptom Score = 20-35)

• Offer treatment options (Guideline.

• Provide the patient with a copy of Treating Your Enlarged Prostate: Patient Guide (Guideline.

• Discuss treatment options, including benefits and risks, after patient reads the Patient Guide (Guideline).

• Moderate symptoms: Wide variation in patient preferences. A substantial number of patients prefer watchful waiting.

• Severe symptoms: More patients prefer surgery to other treatment options. There is still a wide variation in patient preferences overall.

6. Diagnostic Tests: To Confirm BPH Diagnosis (Optional)

• Uroflowmetry, pressure-flow studies, postvoid residual urine volume (PVR), if cause of symptoms is uncertain after the initial evaluation.

• Consider pressure-flow studies if peak flowrate is normal (>15 mL/sec), if patient is at higher risk for a primary bladder problem (known neurologic disease), or in patients for whom a distinction between prostatic obstruction, detrusor instability, and/or impaired detrusor contractility might affect the choice of therapy.

7. Offer Treatment Alternatives

• Educate patient about each treatment option; give him a copy of Treating Your Enlarged Prostate: Patient Guide (Guideline).

• Review benefits and risks of each option.

• Assess patient attitudes toward each treatment option.

• Help patient to assess treatment options and select one.

8. Watchful Waiting

• Periodic (probably annual) reassessment: review symptoms (AUA Symptom Index), physical findings, laboratory testing (Guideline).

• Uroflowmetry and postvoid residual urine determination (Optional).

9. Optional Testing to Plan an Invasive Procedure, Not Used to Determine the Need for Treatment Urethrocystoscopy or transabdominal bladder/prostate ultrasonography to help surgeon plan prostate surgery or balloon dilation by determining prostate size and configuration.

Attachment A: Decision Diagram

Introduction

The number ranges on the Balance Sheet (Attachment B) are 90-percent confidence intervals for the likelihood that a given outcome will follow a given treatment. A wide range indicates that there is considerable uncertainty regarding the likelihood of the outcome.

Explanation of Balance Sheet Table;

Line 1: Likelihood that given patient will experience some symptom improvement; likelihood of improvement greater if pretreatment symptoms more severe. Line 2: Expected amount of improvement (for patients who improve). Line 3: Likelihood that given patient will have treatment complications or adverse events. Line 4: Likelihood that given patient will die due to any causes within 3 months of treatment. Line 5: Likelihood that given patient will experience total incontinence due to the treatment. Line 6: Likelihood that given patient will require surgical correction for a late complication of BPH treatment such as bladder neck contracture or urethral stricture. Line 7: Likelihood that a patient who was potent prior to treatment will experience impotence following treatment. Line 8: Likelihood that a patient who was potent before treatment and still potent after treatment will experience retrograde ejaculation following treatment. Line 9: Estimated number of days a given patient may miss from work during first year of treatment. Line 10: Estimated number of days spent in hospital.

Attachment B. Balance sheet for BPH treatment outcomes

Placebo

• Abrams P, Hollister P, Lawrence J, Doyle PT, Sherwood T, Whitaker RH. Bladder outflow obstruction treated with phenoxybenzamine. Preliminary note. Br J Urol 1982;54:530.

• Abrams PH. Prostatism and prostatectomy: the value of urine flow rate measurement in the preoperative assessment for operation. J Urol 1977;117:70-1.

• Aubrey DA, Khosla T. The effect of 17-alpha-hydroxy-19-norprogesterone caproate (SH 582) on benign prostatic hypertrophy. Br J Surg 1971;58:648-52.

• Beacock CJM, Buck AC, Roberts EE. Bifluranol in the treatment of benign prostatic hyperplasia (BPH). Prostate 1985;7:357-61.

• Becker H, Ebeling L. Konservative therapie der benignen prostata-hyperplasie (BPH) mit Cernilton N. Ergebnisse einer plazebokontrollierten doppelblindstudie. Urologe [B]1988;28:301-6. (Ger).

• Bonard M, de Almeida S, von Niederhausern W. Placebo-controlled double-blind study in human benign obstructive prostatic hypertrophy with flutamide. Eur Urol 1976;2(1):24-8.

• Brooks ME, Sidi AA, Hanani Y, Braf ZF. Ineffectiveness of phenoxybenzamine in treatment of benign prostatic hypertrophy. A controlled study. Urology 1983;21:474-8.

• Buck AC, Cox R, Rees RWM, Ebling L, John A. Treatment of outflow tract obstruction due to benign prostatic hyperplasia with the pollen extract, Cernilton. A double-blind, placebo-controlled study. Br J Urol 1990;66:398-404.

• Caine M, Perlberg S, Gordon R. The treatment of benign prostatic hypertrophy with flutamide (SCH 13521): a placebo-controlled study. J Urol 1975;114:564-8.

• Caine M, Perlberg S, Meretyk S. A placebo-controlled double-blind study of the effect of phenoxybenzamine in benign prostatic obstruction. Br J Urol 1978;50:551-4.

• Carbin B-E, Larsson B, Lindahl O. Treatment of benign prostatic hyperplasia with phytosterols. Br J Urol 1990;66:639-41.

• Castro JE, Griffiths HJL, Edwards DE. A double-blind, controlled, clinical trial of spironolactone for benign prostatic hypertrophy. Br J Surg 1971;58(7):485-9.

• Champault G, Patel JC, Bonnard AM. A double-blind trial of an extract of the plant Serenoa repens in benign prostatic hyperplasia. Br J Clin Pharmacol 1984;18:461-2.

• Chow W, Hahn D, Sandhu D, Slaney P, Henshaw R, Das G, Wells P. Multicentre controlled trial of indoramin in the symptomatic relief of benign prostatic hypertrophy. Br J Urol 1990;65:36-8.

• Donkervoort T, Zinner NR, Sterling AM, Donker PJ, Van Ness J, Ritter RC. Megestrol acetate in treatment of benign prostatic hypertrophy. Urology 1975;6:580-7.

• Ebbinghaus KD, Baur MP. Egrebnisse einer doppelblindstudie ueber die nirksambeit eines medikamentes zur konservativen behandlung des prostata-adenoms. ZFA 1977;53:1054-8. (Ger).

• Farrar DJ. Pryor JS. The effect of bromocriptine in patients with benign prostatic hypertrophy. Br J Urol 1976;48:73-5.

• Geller J, Nelson CG, Albert JD, Pratt C. Effect of megestrol acetate on uroflow rates in patients with benign prostatic hypertrophy: double-blind study. Urology 1979;14:467-74.

• Goodwin MI, Chester JF, Jenkins JD. Acute urinary retention secondary to benign prostatic hyperplasia. Effect of alpha-adrenergic blockade. Urol Int 1986;41:430-1.

• Hedlund H, Andersson K-E, Ek A. Effects of prazosin in patients with benign prostatic obstruction. J Urol 1983;130:275-8.

• Iacovou JW, Dunn M. Indoramin-an effective new drug in the management of bladder outflow obstruction. Br J Urol 1987;60:526-8.

• Jardin A, Bensadoun H, Delauche-Cavallier MC, Attali P, and the BPH-ALF Group. Alfuzosin for treatment of benign prostatic hypertrophy. Lancet 1991;337:1457-61.

• Jensen KM-E, Madsen PO. Candicidin treatment of prostatism: a prospective double-blind placebo-controlled study. Urol Res 1983;11:7-10.

• Kadow C, Abrams PH. A double-blind trial of the effect of beta-sitosteryl glucoside (NA184) in the treatment of benign prostatic hyperplasia. Eur Urol 1986;12:187-9.

• Kawabe K, Ueno A, Takimoto Y, Aso Y, Kato H, and YM617 clinical study group. Use of an alpha 1-blocker, YM617, in the treatment of benign prostatic hypertrophy. J Urol 1990;144:908-12.

• Kirby RS, Coppinger SWC, Corcoran MO, Chapple CR, Flannigan M, Milroy EJG. Prazosin in the treatment of prostatic obstruction. A placebo-controlled study. Br J Urol 1987;60:136-42.

• Klein LA, Lemming B. Balloon dilatation for prostatic obstruction. Long-term follow-up. Urology 1989;33:198-201.

• Lindner A, Siegel YI, Saranga R, Korzcak D, Matzkin H, Braf Z. Complications in hyperthermia treatment of benign prostatic hyperplasia. J Urol 1990;144:1390-2.

• Madsen PO, Dorflinger T, Frimodt-Mller PC, Jensen KM-E. Candicidin in treatment of benign prostatic hypertrophy. J Urol 1984;132:1235-8.

• Martorana G, Giberti C, Damonte P, Ciprandi G, Dirienzo W, Giuliana L. The effect of prazosin in benign prostatic hypertrophy, a placebo controlled double-blind study. IRCS Med Science 1984;12:11-2.

• Matos-Ferreira A, Corte-Real J, Palma J, Durao V. The effect of bromocriptine in benign prostatic hypertrophy and vesicosphincteric dynamics. Br J Urol 1987;60:143-9.

• Orkin L. Efficacy of candicidin in benign prostatic hypertrophy. Urology 1974;4(1):80-4.

• Ostri P, Swartz R, Meyhoff H-H, Petersen JH, Lindgard G, Frimodt-Moller C, Andersson T, Nielsen MS. Antiandrogenic treatment of benign prostatic hyperplasia: a placebo controlled trial. Urol Res 1989;17:29-33.

• Ramsay JW, Scott GI, Whitfield HN. A double-blind controlled trial of a new a-1 blocking drug in the treatment of bladder outflow obstruction. Br J Urol 1985;57:657-9.

• Resnick MI, Jackson JE, Watts LE, Boyce WH. Assessment of the antihypercholesterolemic drug, Probucol, in benign prostatic hyperplasia. J Urol 1983;129:206-9.

• Ronchi F, Margonato A, Ceccardi R, Rigatti P, Rossini BM. Symptomatic treatment of benign prostatic obstruction with nicergoline: a placebo controlled clinical study and urodynamic evaluation. Urol Res 1982;10:131-4.

• Ruutu ML, Hansson E, Juusela HE, Permi JE, Rusk JI, Sotarauta MT, Talja MT, Wuokko EJ, Mattila MJ. Efficacy and side-effects of prazosin as a symptomatic treatment of benign prostatic obstruction. Scand J Urol Nephrol 1991;25:15-9.

• Sertcelik MN, Unal S, Ozgur S, Imamoglu A. Prazosin, a selective alpha, receptor blocker, in the treatment of benign prostatic hypertrophy. Curr Ther Res 1990;48:1066-74.

• Shilcher H, Sauter M, Janssen H. Zum wirksamkeitsnachweis von phytopharmaka am beispeil von prosta-fink. Arztezeitschrift fuer naturheilverfahren. 1985;26(3)177-82. (Ger).

• Smith HR, Memon A, Smart CJ, Dewbury K. The value of permixon in benign prostatic hypertrophy. Br J Urol 1986;58:36-40.

• Sonnenschein R. Untersuchungen der wirksamkeit eines prostatotropen phytotherapeutikums (urtica plus) bei benigner prostatahyperplasie und prostatitis-eine prospektive multizentrische studie. Urologe [B] 1987;27:232-7. (Ger).

• Sporer A, Cohen G, Kamat MH, Seebode JJ. Candicidin physiologic effect on prostate. Urology 1975;6(3):298-304.

• Stone NN. Flutamide in treatment of benign prostatic hypertrophy. Urology (Suppl 4) 1989;34:64-8;discussion 87-96.

• Van Poppel H, Boeckx G, Westlinck KJ, Vereecben RL, Baert L. The efficacy of bromocriptine in benign prostatic hypertrophy. A double-blind study. Br J Urol 1987;60(2):150-2.

• Vontobel, HP, Herzog R, Rutishauser G, Kres H. Ergelnisse einer doppelblinddstudie ueber die wirksamkeit von ERU kapseln in der kinservativen behandlung der benignen prostatahyperplasie. Urologe [A] 1985;24:49-51. (Ger).

Watchful Waiting

• Ball AJ, Feneley RCL, Abrams PH. The natural history of untreated "prostatism." Br J Urol 1981;53:613-6.

• Birkhoff JD, Wiederhorn AR, Hamilton ML, Zinsser HH. Natural history of benign prostatic hypertrophy and acute urinary retention. Urology 1976;7:48-52.

• Clarke R. The prostate and the endocrines: a control series. Br J Urol 1937;9:254-71.

• Craigen AA, Hickling JB, Saunders CR, Carpenter RG. Natural history of prostatic obstruction. J R Coll Gen Pract. 1969;18:226-32.

• Kadow C, Feneley RCL, Abrams PH. Prostatectomy or conservative management in the treatment of benign prostatic hypertrophy? Br J Urol 1988;61:432-4.

Alpha Blockers

• Abrams P, Hollister P, Lawrence J, Doyle PT, Sherwood T, Whitaker RH. Bladder outflow obstruction treated with phenoxybenzamine. Preliminary note. Br J Urol 1982;54:530. Boreham PF, Braithwaite P, Milewski P, Pearson H. Alpha-adrenergic blockers in prostatism. Br J Surg 1977;64:756-7.

• Brooks ME, Sidi AA, Hanani Y, Braf ZF. Ineffectiveness of phenoxybenzamine in treatment of benign prostatic hypertrophy. A controlled study. Urology 1983;21:474-8.

• Caine M, Perlberg S, Meretyk S. A placebo-controlled double-blind study of the effect of phenoxybenzamine in benign prostatic obstruction. Br J Urol 1978;50:551-4.

• Caine M, Perlberg S, Shapiro A. Phenoxybenzamine for benign prostatic obstruction. Review of 200 cases. Urology 1981;17:542-6.

• Chow W, Hahn D, Sandhu D, Slaney P, Henshaw R, Das G, Wells P. Multicentre controlled trial of indoramin in the symptomatic relief of benign prostatic hypertrophy. Br J Urol 1990;65:36-8.

• Dunzendorfer U. Clinical experience: symptomatic management of BPH with terazosin. Urology (6 Suppl) 1988;32:27-31.

• Fabricius PG, Weizert P, Dunzendorfer U, Hannaford JM, Maurath C. Efficacy of once-a-day terazosin in benign prostatic hyperplasia: a randomized, double-blind placebo-controlled clinical trial. Prostate (Suppl) 1990;3:85-93.

• Gerstenberg T, Blaabjerg J, Nielsen ML, Clausen S. Phenoxybenzamine reduces bladder outlet obstruction in benign prostatic hyperplasia. A urodynamic investigation. Invest Urol 1980;18:29-31.

• Goodwin MI, Chester JF, Jenkins JD. Acute urinary retention secondary to benign prostatic hyperplasia. Effect of alpha-adrenergic blockade. Urol Int 1986;41:430-1.

• Hedlund H, Andersson K-E, Ek A. Effects of prazosin in patients with benign prostatic obstruction. J Urol 1983;130:275-8.

• Iacovou JW, Dunn M. Indoramin-an effective new drug in the management of bladder outflow obstruction. Br J Urol 1987;60:526-8.

• Jardin A, Bensadoun H, Delauche-Cavallier MC, Attali P, and the BPH-ALF Group. Alfuzosin for treatment of benign prostatic hypertrophy. Lancet 1991;337:1457-61.

• Kawabe K, Niijima T. Use of an alpha 1-blocker, YM-12617, in micturition difficulty. Urol Int 1987;42:280-4.

• Kawabe K, Ueno A, Takimoto Y, Aso Y, Kato H, and YM617 clinical study group. Use of an alpha 1-blocker, YM617, in the treatment of benign prostatic hypertrophy. J Urol 1990;144:908-12.

• Kirby RS, Coppinger SWC, Corcoran MO, Chapple CR, Flannigan M, Milroy EJG. Prazosin in the treatment of prostatic obstruction. A placebo-controlled study. Br J Urol 1987;60:136-42.

• Lepor H, Knapp-Maloney G, Sunshine H. A dose titration study evaluating terazosin, a selective, once-a-day alpha-1-blocker for the treatment of symptomatic benign prostatic hyperplasia. J Urol 1990;144:1393-8.

• Lepor H, Henry D, Laddu AR. The efficacy and safety of terazosin for the treatment of symptomatic BPH. Prostate 1991;18:345-55.

• Lepor H, Knapp-Maloney G. Outcome assessment of terazosin for benign prostatic hyperplasia (BPH): 18 month follow-up. Abstract #204. J Urol (Suppl) 1991;145:263A.

• Martorana G, Giberti C, Damonte P, Ciprandi G, Dirienzo W, Giuliana L. The effect of prazosin in benign prostatic hypertrophy, a placebo controlled double-blind study. IRCS Med Science 1984;12:11-2.

• Ramsay JW, Scott GI, Whitfield HN. A double-blind controlled trial of a new a-1 blocking drug in the treatment of bladder outflow obstruction. Br J Urol 1985;57:657-9.

• Ronchi F, Margonato A, Ceccardi R, Rigatti P, Rossini BM. Symptomatic treatment of benign prostatic obstruction with nicergoline: a placebo controlled clinical study and urodynamic evaluation. Urol Res 1982;10:131-4.

• Ruutu ML, Hansson E, Juusela HE, Permi JE, Rusk JI, Sotarauta MT, Talja MT, Wuokko EJ, Mattila MJ. Efficacy and side-effects of prazosin as a symptomatic treatment of benign prostatic obstruction. Scand J Urol Nephrol 1991;25:15-9.

• Sertcelik MN, Unal S, Ozgur S, Imamoglu A. Prazosin, a selective alpha, receptor blocker, in the treatment of benign prostatic hypertrophy. Curr Ther Res 1990;48:1066-74.

Finasteride

The BPH panel performed the analysis on data submitted to the Food and Drug Administration that have subsequently been published in the following papers:

• Gormley GJ, Stoner E, Bruskewitz RC, Imperato-McGinley J, Walsh PC, McConnell JD, Andriole GL, Geller J, Bracken BR, Tenover JS, Vaughan ED, Pappas F, Taylor A, Binkowitz B, Ng J, et al. The effect of finasteride in men with benign prostatic hyperplasia. N Engl J Med 1992;327:1185-91.

• Finasteride Study Group. Finasteride (MK-906) in the treatment of benign prostatic hyperplasia. Prostate 1993;22:291-9.

Balloon Dilation

• Castaneda F, Reddy P, Wasserman N, Hulbert J, Lund G, Letourneau JG, Hunter DW, Castaneda-Zuniga WR, Amplatz K. Benign prostatic hypertrophy: retrograde transurethral dilation of the prostatic urethra in humans. Work in progress. Radiology 1987;163:649-53.

• Daughtry JD, Rodan BA, Bean WJ. Balloon dilation of prostatic urethra. Urology 1990;36:203-9.

• Gill KP, Machan LS, Allison DJ, Williams G. Bladder outflow tract obstruction and urinary retention from benign prostatic hypertrophy treated by balloon dilatation. Br J Urol 1989;64:618-22.

• Goldenberg SL, Perez-Marrero RA, Lee LM, Emerson L. Endoscopic balloon dilation of the prostate: early experience. J Urol 1990;144:83-9.

• Keane PF, Charig CR, Hudd C, Shah PJR, Kellett MJ, Boyle J, Wickham JEA, O'Donoghue EPN. Balloon dilatation of the prostate: technique and early results. Br J Urol 1990;65:354-6.

• Klein L, Perez-Marrero R, Bowers GW, Ludwig JJ, Herwig K. Transurethral cystoscopic balloon dilation of the prostate. J Endourol 1990;4(2):183-91.

• Klein LA, Lemming B. Balloon dilatation for prostatic obstruction. Long-term follow-up. Urology 1989;33:198-201.

• Lepor H, Sypherd D, Machi G, Derus J. Randomized double-blind study comparing the effectiveness of balloon dilation of the prostate and cystoscopy for the treatment of symptomatic benign prostatic hyperplasia. J Urol 1992;147(3):639-44.

• McLoughlin J, Keane PF, Jager R, Gill KP, Machann L, Williams G. Dilatation of the prostatic urethra with 35 mm balloon. Br J Urol 1991;67:177-81.

• Wasserman NF, Reddy PK, Zhang G, Berg PA. Experimental treatment of benign prostatic hyperplasia with transurethral balloon dilation of the prostate: preliminary study of 73 humans. Radiology 1990;177:485-94.

TUIP

• D'Ancona CAL, Netto NR Jr, Cara AM, Ikari O. Internal urethrotomy of the prostatic urethra or transurethral resection in benign prostatic hyperplasia. J Urol 1990;144:918-20.

• Delaere KPJ, Debruyne FMJ, Moonen WA. Extended bladder neck incision for outflow obstruction in male patients. Br J Urol 1983;55:225-8.

• Dorflinger T, Oster M, Larsen JF, Walter S, Krarup T. Transurethral prostatectomy or incision of the prostate in the treatment of prostatism caused by small benign prostates. Scand J Urol Nephrol (Suppl) 1987;104:77-81.

• Edwards LE, Bucknall TE, Pittam MR, Richardson DR, Stanek J. Transurethral resection of the prostate and bladder neck incision: a review of 700 cases. Br J Urol 1985;57:168-71.

• Hedlund H, Ek A. Ejaculation and sexual function after endoscopic bladder neck incision. Br J Urol 1985;57:164-7.

• HellstrOm P, Lukkarinen O, Kontturi M. Bladder neck incision or transurethral electroresection for the treatment of urinary obstruction caused by a small benign prostate? A randomized urodynamic study. Scand J Urol Nephrol 1986;20:187-92.

• Katz PG, Greenstein A, Ratliff JE, Marks S, Guice J. Transurethral incision of the bladder neck and prostate. J Urol 1990;144:694-6.

• Kelly MJ, Roskamp D, Leach GE. Transurethral incision of the prostate: a preoperative and postoperative analysis of symptoms and urodynamic findings. J Urol 1989;142:1507-9.

• Larsen EH, Dorflinger T, Gasser TC, Graversen PH, Bruskewitz RC. Transurethral incision versus transurethral resection of the prostate for the treatment of benign prostatic hypertrophy. A preliminary report. Scand J Urol Nephrol (Suppl) 1987;104:83-6.

• Li MK, Ng ASM. Bladder neck resection and transurethral resection of the prostate: a randomized prospective trial. J Urol 1987;138:807-9.

• Mobb GE, Moisey CU. Long-term follow-up of unilateral bladder neck incision. Br J Urol 1988;62:160-2.

• Nielsen HO. Transurethral prostatotomy versus transurethral prostatectomy in benign prostatic hypertrophy. A prospective randomised study. Br J Urol 1988;61:435-8.

• Orandi A. Transurethral incision of prostate (TUIP): 646 cases in 15 years -- a chronological appraisal. Br J Urol 1985;57:703-7.

• Orandi A. Transurethral incision of prostate compared with transurethral resection of prostate in 132 matching cases. J Urol 1987;138:810-5.

• Orandi A. Transurethral resection versus transurethral incision of the prostate. Urol Clin North Am 1990;17:601-12.

TURP

• Aalkjer V. Transurethral resection/prostatectomy versus dilatation treatment in hypertrophy of the prostate II. Urol Int 1965;20:17-22. Abrams PH, Farrar DJ, Turner-Warwick RT, Whiteside CG, Feneley RC. The results of prostatectomy: a symptomatic and urodynamic analysis of 152 patients. J Urol 1979;121:640-2.

• Ball AJ, Smith PJB. The long-term effects of prostatectomy: a uroflowmetric analysis. J Urol 1982;128:538-40.

• Bandhauer K. Transurethral prostatectomy-complications. N Dev Biosc 1989;5:81-7.

• Bruskewitz RC, Larsen EH, Madsen PO, Dorflinger T. 3-year followup of urinary symptoms after transurethral resection of the prostate. J Urol 1986;136:613-5.

• Chilton CP, Morgan RJ, England HR, Paris AMI, Blandy JP. A critical evaluation of the results of transurethral resection of the prostate. Br J Urol 1978;50:542-6.

• D'Ancona CAL, Netto NR Jr, Cara AM, Ikari O. Internal urethrotomy of the prostatic urethra or transurethral resection in benign prostatic hyperplasia. J Urol 1990;144:918-20.

• Dorflinger T, Oster M, Larsen JF, Walter S, Krarup T. Transurethral prostatectomy or incision of the prostate in the treatment of prostatism caused by small benign prostates. Scand J Urol Nephrol (Suppl) 1987;104:77-81.

• Edwards LE, Bucknall TE, Pittam MR, Richardson DR, Stanek J. Transurethral resection of the prostate and bladder neck incision: a review of 700 cases. Br J Urol 1985;57:168-71.

• HellstrOm P, Lukkarinen O, Kontturi M. Bladder neck incision or transurethral electroresection for the treatment of urinary obstruction caused by a small benign prostate? A randomized urodynamic study. Scand J Urol Nephrol 1986;20:187-92.

• Holtgrewe HL, Valk WL. Late results of transurethral prostatectomy. J Urol 1964;92:51-5.

• Kadow C, Feneley RCL, Abrams PH. Prostatectomy or conservative management in the treatment of benign prostatic hypertrophy? Br J Urol 1988;61:432-4.

• Larsen EH, Dorflinger T, Gasser TC, Graversen PH, Bruskewitz RC. Transurethral incision versus transurethral resection of the prostate for the treatment of benign prostatic hypertrophy. A preliminary report. Scand J Urol Nephrol (Suppl) 1987;104:83-6.

• Lentz HC, Mebust WK, Foret JD, Melchior J. Urethral strictures following transurethral prostatectomy: review of 2,223 resections. J Urol 1977;117:194-6.

• Lepor H, Rigaud G. The efficacy of transurethral resection of the prostate in men with moderate symptoms of prostatism. J Urol 1990;143(3):533-7.

• Li MK, Ng ASM. Bladder neck resection and transurethral resection of the prostate: a randomized prospective trial. J Urol 1987;138:807-9.

• Lund B-L, Dingsor E. Benign obstructive prostatic enlargement. A comparison between the results of treatment by transurethral electro-resection and the results of open surgery. Scand J Urol Nephrol 1976;10:33-8.

• Malone PR, Cook A, Edmonson R, Gill MW, Shearer RJ. Prostatectomy: patients' perception of long-term follow up. Br J Urol 1988;61:234-8.

• Mebust WK, Holtgrewe HL, Cockett ATK, Peters PC, and Writing Committee. Transurethral prostatectomy: immediate and postoperative complications. A cooperative study of 13 participating institutions evaluating 3,885 patients. J Urol 1989;141:243-7.

• Melchior J, Valk WL, Foret JD, Mebust WK. Transurethral prostatectomy: computerized analysis of 2,223 consecutive cases. J Urol 1974a;112:634-42.

• Melchior J, Valk WL, Foret JD, Mebust WK. Transurethral prostatectomy in the azotemic patient. J Urol 1974b;112:643-6.

• Meyhoff HH, Nordling J. Long term results of transurethral and transvesical prostatectomy. A randomized study. Scand J Urol Nephrol 1986;20:27-33.

• Meyhoff HH, Nordling J, Hald T. Urodynamic evaluation of transurethral versus transvesical prostatectomy. A randomized study. Scand J Urol Nephrol 1984;18:27-35.

• Meyhoff HH, Nordling J, Hald T. Economy in transurethral prostatectomy. Scand J Urol Nephrol 1985;19:17-20.

• Nielsen HO. Transurethral prostatotomy versus transurethral prostatectomy in benign prostatic hypertrophy. A prospective randomised study. Br J Urol 1988;61:435-8.

• Orandi A. Transurethral incision of prostate compared with transurethral resection of prostate in 132 matching cases. J Urol 1987;138:810-5.

• Orandi A. Transurethral resection versus transurethral incision of the prostate. Urol Clin North Am 1990;17:601-12.

• Roos NP, Ramsey EW. A population-based study of prostatectomy: outcomes associated with differing surgical approaches. J Urol 1987;137:1184-8.

• Roos NP, Wennberg JE, Malenka DJ, Fisher ES, McPherson K, Andersen TF, Cohen MM, Ramsey E. Mortality and reoperation after open and transurethral resection of the prostate for benign prostatic hyperplasia. N Engl J Med 1989;320:1120-4.

• Singh M, Tresidder GC, Blandy JP. The evaluation of transurethral resection for benign enlargement of the prostate. Br J Urol 1973;45:93-102.

Open Surgery

• Arvola I, Lilius HG. Transvesical prostatectomy a.m. Hryntschak. Ann Chir Gynaecol Fenn 1972;61:74-8.

• Ball AJ, Smith PJB. The long-term effects of prostatectomy: a uroflowmetric analysis. J Urol 1982;128:538-40.

• Beck AD. The Hryntschak prostatectomy II: a late review of 179 cases. J Urol 1970;103:778-82.

• Beck AD, Gaudin HJ. The Hryntschak prostatectomy I: a review of 1,346 cases. J Urol 1970;103:637-40.

• Blue GD, Campbell JM. A clinical review of one thousand consecutive cases of retropubic prostatectomy. J Urol 1958;80:257-9.

• Bollmann J, Zingg E. Retropubic prostatectomy. Prog Clin Biol Res 1976;6:59-65.

• Davillas NE, Miliaresis A, Katsoulis A, Katatigiotis S. Observations on 1,000 Millin prostatectomies. Eur Urol 1978;4:100-2.

• Gregoir W. Haemostatic prostatic adenomectomy. Eur Urol 1978;4:1-8.

• Hohenfellner R. Suprapubic prostatectomy. In: Prostatic diseases. New York: Alan R. Liss, Inc.;1976. p. 49-57.

• Lenko J. Millin's retropubic prostatectomy. A clinical study. Int Urol Nephrol 1977;9:25-32.

• Lenko J, Cieslinski S. Millin's retropubic prostatectomy: report of 233 cases. Br J Urol 1965;37:450-4.

• Lesiewicz H, Cieslinski S. Millin's retropubic prostatectomy: a clinical study. Int Urol Nephrol 1985;17:341-8.

• Lich R. Retropubic prostatectomy: a review of 678 patients. J Urol 1954;72:434-8.

• Lund B-L, Dingsor E. Benign obstructive prostatic enlargement. A comparison between the results of treatment by transurethral electro-resection and the results of open surgery. Scand J Urol Nephrol 1976;10:33-8.

• Macky W. Hryntschak prostatectomy: experiences with 300 consecutive cases. Br J Urol 1961;33:19-23.

• Magasi P, Vegh A. One thousand cases of prostatectomy. Acta Chir Hung 1984;25:63-74.

• Meyhoff HH, Nordling J. Long term results of transurethral and transvesical prostatectomy. A randomized study. Scand J Urol Nephrol 1986;20:27-33.

• Meyhoff HH, Nordling J, Hald T. Urodynamic evaluation of transurethral versus transvesical prostatectomy. A randomized study. Scand J Urol Nephrol 1984;18:27-35.

• Meyhoff HH, Nordling J, Hald T. Economy in transurethral prostatectomy. Scand J Urol Nephrol 1985;19:17-20.

• Nicoll GA, Riffle GN, Anderson FO. Suprapubic prostatectomy. The removable purse string: a continuing comparative analysis of 300 consecutive cases. J Urol 1978;120:702-4.

• Roos NP, Ramsey EW. A population-based study of prostatectomy: outcomes associated with differing surgical approaches. J Urol 1987;137:1184-8.

• Roos NP, Wennberg JE, Malenka DJ, Fisher ES, McPherson K, Andersen TF, Cohen MM, Ramsey E. Mortality and reoperation after open and transurethral resection of the prostate for benign prostatic hyperplasia. N Engl J Med 1989;320:1120-4.

• Salvaris M. Retropubic prostatectomy: an evaluation of 1200 operations. Med J Aust 1969;1:370-6.

• Singh M, Tresidder GC, Blandy JP. The evaluation of transurethral resection for benign enlargement of the prostate. Br J Urol 1973;45:93-102.

• Stearns DB. Retropubic prostatectomy, 1947-1960: A critical evaluation. J Urol 1961;85:322-8.

• Taylor WN, Kaylor WM, Taylor JN. Retropubic and suprapubic prostatectomy: comparative clinical study. J Urol 1955;74:129-37.

• Taylor Z, Krakauer H. Mortality and reoperation following prostatectomy: outcomes in a medicare population. Urology (Suppl 1) 1991;38:27-31.